Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
New and Bestselling from
Wiley-IEEE Press
A Guide to the Wireless
Engineering Body of Knowledge
(WEBOK)
The ComSoc Guide to Next
Generation Optical Transport
IEEE COMMUNICATIONS SOCIETY
The ultimate reference book for professionals
in the wireless industry and study guide for the
WCET. The information presented in this book
reflects the evolution of wireless technologies,
their impact on the profession, and the industry’s
commonly accepted best practices.
HUUB VAN HELVOORT
Provides a unique overview of SDH and OTN for
engineers who are new to the field, as well as
manufacturers, network operators, and graduate
students who need a basic understanding of the
topics.
978-0-470-43366-9 • April 2009 • Pbk • 272pp
$69.95
SDH/SONET/OTN
978-0-470-22610-0 • October 2009 • Pbk • 211pp
$63.50
ComSoc Guides to Communications
Technologies
Handbook on Array Processing and
Sensor Networks
Wireless Sensor Networks
SIMON HAYKIN and K. J. RAY LIU
Provides readers with a collection of tutorial articles
contributed by world-renowned experts on recent
advancements and the state of the art in array processing
and sensor networks.
JUN ZHENG and ABBAS JAMALIPOUR
This book provides a comprehensive and systematic
introduction to the fundamental concepts, major
challenges, and effective solutions in wireless sensor
networking (WSN).
978-0-470-37176-3 • January 2010 • Hbk • 904pp
$185.00
Adaptive and Learning Systems for Signal Processing, Communications
and Control Series
978-0-470-16763-2 • October 2009 • Hbk • 489pp
$94.95
A Networking Perspective
Ground-Based Wireless Positioning
Advances in Multiuser Detection
MICHAEL L. HONIG
During the past decade, the design and development of
current and emerging wireless systems have motivated
many important advances in multiuser detection. This
book provides a comprehensive overview of crucial recent
developments
978-0-470-47381-8 • September 2009 • Hbk • 493pp
$125.00
Wiley Series in Telecommunications and Signal
Processing
Next Generation Solutions
TULAY ADALI and SIMON HAYKIN
Recent developments have made it clear that significant
performance gains can be achieved beyond those using
standard adaptive filtering approaches. This book presents
the next generation of algorithms that will produce these
desired results.
978-0-470-19517-8 • April 2010 • Hbk • 424pp
$120.00
Adaptive and Learning Systems for Signal Processing, Communications
and Control Series
IEEE
Wireless Positioning
Kegen Yu Ian Sharp Y. Jay Guo
978-0-470-74704-9 • June 2009 • Hbk • 450pp
$120.00
Near-Capacity Multi-Functional
MIMO Systems
LAJOS HANZO, OSAMAH ALAMRI, MOHAMMED
EL-HAJJAR and NAN WU
Providing an all-encompassing self-contained treatment
of Near-Capacity Multi-Functional MIMO Systems it
gives a detailed examination of wireless landscape,
including the fields of channel coding, spacetime coding and turbo
detection techniques.
978-0-470-77965-1 • May 2009 • Hbk • 738pp
$200.00
ORDER WILEY ONLINE…
For telephone orders:
Call 1-800-225-5945 (1-800-US-WILEY)
Fax: (212) 850-8888
_________
E-mail: custserv@wiley.com
• Visit the Wiley Electrical Engineering homepage: www.wiley.com/electrical
• Wiley Communications Technology Website: www.wiley.com/go/commstech/
• Amazon Wiley Communications Technology bookstore: www.amazon.com/professional/
Please note all prices correct at time of going to press but subject to change.
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
14140
HOW TO ORDER
Communications
Ground-Based
Sphere-Packing, Iterative Detection and
Cooperation
Adaptive Signal Processing
Wiley books are available through your Bookseller.
Alternatively send your order to:
John Wiley & Sons Inc., 111 River Street, Hoboken,
NJ 07030, USA
KEGEN YU, IAN SHARP and Y. JAY GUO
Provides an in-depth treatment of non-GPS based
wireless positioning techniques, with a balance between
theory and engineering practice. The book presents the
architecture, design and testing of a variety of wireless
positioning systems based on the time-of-arrival, signal
strength, and angle-of-arrival measurements.
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
THIS MONTH’S DIGITAL DELIVERY OF
IEEE COMMUNICATIONS MAGAZINE
SUPPORTED BY:
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Contents | Zoom in | Zoom out
For navigation instructions please click here
IEEE
Search Issue | Next Page
April 2010, Vol. 48, No. 4
www.comsoc.org
C
FT
D
l:
ria
to
Tu 9
c ge
So Pa
om ee
S
Integrated Circuits for
Communications
ee
Fr
Design and Implementation:
IMS Applications and Support
MAGAZINE
®
A Publication of the IEEE Communications Society
Contents | Zoom in | Zoom out
For navigation instructions please click here
Search Issue | Next Page
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
©2009 Cisco Systems Inc. All Rights Reserved.
IEEE
Now Showing:
The Future of Video Entertainment
Video is driving a market transition in networking
that creates unique challenges for service providers.
Evolve your network with Cisco into a medianet
and go beyond being a service provider
to become an experience provider.
Cisco is the only company, which along with its partners,
can address video and rich media
from an end-to-end perspective
by leveraging deep expertise in IP, video
and customer premise solutions.
Move video more efficiently,
and energize customer satisfaction
and increase revenues.
Deliver the connected life that’s more social,
more personal and more interactive.
Begin your transformation at www.cisco.com/go/sp/medianet
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Accelerating Growth and Development
COMMUNICATIONS:
____________________
OPENING & KEYNOTE SESSIONS
Pandor
Gray
Pandey
September
Accelerating Growth and Development
Monday, 24 May 2010 • 8:30 – 9:40
Ngcaba
Accelerating Growth and
Development for Business
Monday, 24 May 2010 • 10:15 – 11:30
Valenzuela
Yeung
Research and Technology for
Accelerating Growth and Development
Tuesday, 25 May 2010 • 8:30 – 10:00
Naledi Pandor
Reuben September
Reinaldo Valenzuela
Conference Chair
Minister of the Department
of Science and Technology
South Africa
Group Chief Executive Officer
Telkom, Republic of South Africa
Director, Wireless Research
Bell Labs, Alcatal-Lucent, USA
Steven D. Gray
Head & Vice President, Corporate Research
Huawei Technologies, USA
Ajay Pandey
Raymond W. Yeung
Managing Director & Chief Executive Officer
Neotel, Republic of South Africa
Chair Professor
Chinese University of Hong Kong, Hong Kong
Andile Ngcaba
Executive Chairman Dimension Data - Africa
and Middle East, Republic of South Africa
BUSINESS FORUMS
Monday, 24 March 2010
Tuesday, 25 March 2010
Wednesday, 26 March 2010
Mobile Broadband as an Enabler of
Accelerated Broadband Development
Regulatory Environment and Spectrum Reform
– the Key to Growth and Development
Ultra Broadband Technologies
13:30 – 15:00
11:00 – 12:30
9:00 – 10:30
IMT-Advanced - The Technical and the Strategic
Landscape for 4G and Beyond
Policy and Funding Models for Accelerating the
Rollout of Broadband Access and Networks in
Developing Countries
100 Gbps Networking: Trends, Deployment, and
Challenges
15:30 – 17:00
Alternatives to providing Affordable and Sustainable
Mobile Broadband in a Fast and Cost Effective
Manner
13:30 – 15:00
Spectrum Management – Technologies, Economics,
Regulatory Environment
15:30 – 17:00
Challenges in Commercialization of Cognitive Radio
Technologies
TECHNICAL SYMPOSIA
• Ad Hoc Sensor and Mesh Networking
• Communication and Information System Security
• Communication Theory
• Communications Quality of Service, Reliability and
Performance Modeling
• Multimedia Services, Communications Software
and Services
• Next Generation Networking Protocols, and
Services
• Optical Networks and Systems
TUTORIALS
• Signal Processing
• Wireless Communications
• Wireless and Mobile Networking
• Selected Areas in Communications
> Buy one get the second one free! <
T1:
T2:
T3:
T4:
T5:
T6:
T7:
T8:
T9:
Cooperative Wireless Communications
Networking Cognitive Radios for Dynamic Spectrum Access
Broadband Wireless Technologies: LTE and WiMax
Locality Aware P2P Delivery: The Way to Scale Internet Video
Energy Efficient Networks
High-Definition Location-Awareness
Compressive Sensing and Signal Scarcity in WirelessCommunications
Application of Game Theory for Designing Cognitive Radio Networks
Statistical Delay-QoS Provisioning in Wireless Networks: Effective Capacity
and QoS-Driven Resource Allocations
T10: Planning Wireless Municipal Networks based on Wi-Fi/WiMax Mesh Networks Applications, Technologies and Business Models
T11:
T12:
T13:
T14:
T15:
T16:
T17:
T18:
T19:
T20:
Femto-cells: Opportunities and Challenges
Vehicular Networking
Aspects of Multiuser MIMO-Principles and Standardization in LTE-Advanced
Understanding Next Generation Mobile Networks (NGMN): The Role of the
Evolved Packet Core (EPC) for seamless Mobile Broadband Service Provision
Multi Gigabit Transmissions at 60 GHz: Standards, Technologies, and Challenges
Stochastic Geometry and Random Graphs for the Analysis and Design of
Wireless Networks
Recovery in IP over Optical Networks: Challenges and Solutions
Overview of 3GPP LTE Radio Interface: Layers 2/3
Security Issues in Dynamic Spectrum Access Networks
Biologically-inspired and Nano-scale Communication and Networking
WORKSHOPS
W1: Smart Grid Communications
W2: Energy Efficiency in Wireless Networks & Wireless Networks for Energy
Efficiency
W3: Vehicular Connectivity
W4: Social Networks
W5: Underwater Networks
W6:
W7:
W8:
W9:
Cooperative and Cognitive Mobile Networks
Cognitive Radio Interfaces and Signal Processing
Medical Applications Networking
IEEE Vehicular Networking and Applications: Future Wireless
Technologies for Vehicle Infrastructure Integration (VII) Applications
W10: Integrated Disaster Risk Management for Africa
Register by 23 April 2010 to receive the reduced registration rate!
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Director of Magazines
Andrzej Jajszczyk, AGH U. of Sci. & Tech. (Poland)
A
BEMaGS
F
IEEE
Editor-in-Chief
Steve Gorshe, PMC-Sierra, Inc. (USA)
Associate Editor-in-Chief
Sean Moore, Avaya (USA)
Senior Technical Editors
Tom Chen, Swansea University (UK)
Nim Cheung, ASTRI (China)
Nelson Fonseca, State Univ. of Campinas (Brazil)
Torleiv Maseng, Norwegian Def. Res. Est. (Norway)
Peter T. S. Yum, The Chinese U. Hong Kong (China)
Technical Editors
Sonia Aissa, Univ. of Quebec (Canada)
Mohammed Atiquzzaman, U. of Oklahoma (USA)
Paolo Bellavista, DEIS (Italy)
Tee-Hiang Cheng, Nanyang Tech. U. (Rep. Singapore)
Jacek Chrostowski, Scheelite Techn. LLC (USA)
Sudhir S. Dixit, Nokia Siemens Networks (USA)
Stefano Galli, Panasonic R&D Co. of America (USA)
Joan Garcia-Haro, Poly. U. of Cartagena (Spain)
Vimal K. Khanna, mCalibre Technologies (India)
Janusz Konrad, Boston University (USA)
Abbas Jamalipour, U. of Sydney (Australia)
Deep Medhi, Univ. of Missouri-Kansas City (USA)
Nader F. Mir, San Jose State Univ. (USA)
Amitabh Mishra, Johns Hopkins University (USA)
Sedat Ölçer, IBM (Switzerland)
Glenn Parsons, Ericsson Canada (Canada)
Harry Rudin, IBM Zurich Res.Lab. (Switzerland)
Hady Salloum, Stevens Institute of Tech. (USA)
Antonio Sánchez Esguevillas, Telefonica (Spain)
Heinrich J. Stüttgen, NEC Europe Ltd. (Germany)
Dan Keun Sung, Korea Adv. Inst. Sci. & Tech. (Korea)
Danny Tsang, Hong Kong U. of Sci. & Tech. (Japan)
Series Editors
Ad Hoc and Sensor Networks
Edoardo Biagioni, U. of Hawaii, Manoa (USA)
Silvia Giordano, Univ. of App. Sci. (Switzerland)
Automotive Networking and Applications
Wai Chen, Telcordia Technologies, Inc (USA)
Luca Delgrossi, Mercedes-Benz R&D N.A. (USA)
Timo Kosch, BMW Group (Germany)
Tadao Saito, University of Tokyo (Japan)
Design & Implementation
Sean Moore, Avaya (USA)
Integrated Circuits for Communications
Charles Chien (USA)
Zhiwei Xu, SST Communication Inc. (USA)
Stephen Molloy, Qualcomm (USA)
Network and Service Management Series
George Pavlou, U. of Surrey (UK)
Aiko Pras, U. of Twente (The Netherlands)
Topics in Optical Communications
Hideo Kuwahara, Fujitsu Laboratories, Ltd. (Japan)
Osman Gebizlioglu, Telcordia Technologies (USA)
John Spencer, Optelian (USA)
Vijay Jain, Verizon (USA)
Topics in Radio Communications
Joseph B. Evans, U. of Kansas (USA)
Zoran Zvonar, MediaTek (USA)
Standards
Yoichi Maeda, NTT Adv. Tech. Corp. (Japan)
Mostafa Hashem Sherif, AT&T (USA)
Columns
Book Reviews
Andrzej Jajszczyk, AGH U. of Sci. & Tech. (Poland)
History of Communications
Mischa Schwartz, Columbia U. (USA)
Regulatory and Policy Issues
J. Scott Marcus, WIK (Germany)
Jon M. Peha, Carnegie Mellon U. (USA)
Technology Leaders' Forum
Steve Weinstein (USA)
Very Large Projects
Ken Young, Telcordia Technologies (USA)
Your Internet Connection
Eddie Rabinovitch, ECI Technology (USA)
Publications Staff
Joseph Milizzo, Assistant Publisher
Eric Levine, Associate Publisher
Susan Lange, Digital Production Manager
Jennifer Porcello, Production Specialist
Catherine Kemelmacher, Associate Editor
Devika Mittra, Publications Assistant
®
2
Communications
IEEE
MAGAZINE
April 2010, Vol. 48, No. 4
www.comsoc.org/~ci
6
10
14
16
19
PRESIDENT’S PAGE/BYEONG GI LEE
SOCIETY NEWS/COMSOC MEMBERS NAMED TO FELLOW GRADE
CONFERENCE CALENDAR
NEW PRODUCTS
GLOBAL COMMUNICATIONS NEWSLETTER/EDITED BY STEFANO BREGNI
DESIGN AND IMPLEMENTATION: IMS APPLICATIONS AND SUPPORT
SERIES EDITOR: SEAN MOORE
24 GUEST EDITORIAL
26 IMS SERVICE DEVELOPMENT API AND TESTBED
The authors explore the architectural and protocol aspects that enable third-party
IMS application development and deployment. They also study how the applications
will coexist with other applications already deployed in the IMS. They describe Java
libraries exploiting the functionality of the IMS both in the terminal client and
within the core network.
SALVATORE LORETO, TOMAS MECKLIN, MILJENKO OPSENICA, AND HEIDI-MARIA RISSANEN
34 DEPLOYMENT OF CONTEXTUAL CORPORATE TELCO SERVICES BASED ON
PROTOCOL ADAPTATION IN THE NGN ENVIRONMENT
The authors present a practical deployment of a contextual service offered by a
convergent telecommunications operator, whose functionality is to provide
intelligent context-based call routing and rerouting, orchestrated from the operator’s
service layer. It is based on IMS control layer capabilities to properly capture the
situation of the end user in a ubiquitous coverage area.
ALEJANDRO CADENAS, ANTONIO SANCHEZ-ESGUEVILLAS, AND BELÉN CARRO
42 THE DESIGN AND IMPLEMENTATION OF ARCHITECTURAL COMPONENTS FOR
THE INTEGRATION OF THE IP MULTIMEDIA SUBSYSTEM AND WIRELESS
SENSOR NETWORKS
The authors have previously proposed a presence-based architecture for WSN/IMS
integration. This architecture relies on two key components: a WSN/IMS gateway
acting as an interworking unit between WSNs and the IMS; and an extended presence
server serving as a context information management node in the core network. In
this article the authors focus on the design and implementation of these two
components.
MAY EL BARACHI, ARIF KADIWAL, ROCH GLITHO, FERHAT KHENDEK, RACHIDA DSSOULI
52 BROADBAND INTERNET IN EU COUNTRIES: LIMITS TO GROWTH
The author provides an analysis of broadband Internet diffusion in 27 countries of
the European Union. The author also proposes a simple model of its growth and
identifies the theoretical growth limits in each country. Some aspects of the
European i2010 project implementation are presented, discussed, and compared
with the model.
RYSZARD STRUZAK
58 SERVICE TRAFFIC MANAGEMENT SYSTEM FOR MULTISERVICE IP NETWORKS:
LESSONS LEARNED AND APPLICATIONS
Next-generation networks offer new opportunities and challenges to Internet service
providers as well as providers of other online services. Service providers can now
deploy new services over an IP network infrastructure without building their own
networks. In an open network environment, the network resources of ISPs should
be fairly open to third parties that plan to launch their own services. To respond to
the changing network paradigm, it is essential to measure the traffic of individual
services, and to estimate their cost for cost accounting between service provider and
ISP. However, current traffic measurement techniques only provide total traffic
volume in links, without reporting the operator whose services flow through the links.
JUNGYUL CHOI, SEUNG-HOON KWAK, MI-JEONG LIM, TAEIL CHAE, BYOUNG-KWON SHIM, AND
JAE-HYOUNG YOO
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
“I need a high performance signal analyzer that will take
me well into my wireless future.”
How about 4G and beyond?
The problem with time is it never stops moving forward. Imagine a high
performance signal analyzer that delivers seamless integration now, but
also over time to maximize flexibility, scalability and bench top longevity.
Meet the new Agilent PXA signal analyzer, just one of the test tools in
the X-Series, an evolutionary approach to signal analysis that spans
instruments, measurements and software. With upgradable hardware
including CPU, hard drives, I/O, memory, and expansion slots, it’s ready
to evolve with you starting now.
PXA Signal Analyzer (N9030A)
Up to 75 dB of SFDR* over 140 MHz analysis bandwidth for multi-carrier analysis
Up to -87 dBc 3GPP ACLR dynamic range
LTE and HSPA+ are just two of numerous one-button measurement applications
Agilent 89600 VSA software for deep capture and analysis
Save up to 40% on measurement applications and
view LTE ACLR, W-CDMA, Noise Floor Extension
app notes at www.agilent.com/find/beyond
*spurious free dynamic range
© 2010 Agilent Technologies
Communications
IEEE
u.s. 1-800-829-4444
canada: 1-877-894-4414
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
2010 Communications Society
Elected Officers
Byeong Gi Lee, President
Doug Zuckerman, Past President
Mark Karol, VP–Technical Activities
Khaled B. Letaief, VP–Conferences
Sergio Benedetto, VP–Member Relations
Leonard Cimini, VP–Publications
Members-at-Large
Class of 2010
Fred Bauer, Victor Frost
Stefano Galli, Lajos Hanzo
Class of 2011
Robert Fish, Joseph Evans
Nelson Fonseca, Michele Zorzi
Class of 2012
Stefano Bregni, V. Chan
Iwao Sasase, Sarah K. Wilson
2010 IEEE Officers
Pedro A. Ray, President
Moshe Kam, President-Elect
David G. Green, Secretary
Peter Staecker, Treasurer
John R. Vig, Past-President
E. James Prendergast, Executive Director
Nim Cheung, Director, Division III
IEEE COMMUNICATIONS MAGAZINE (ISSN 01636804) is published monthly by The Institute of Electrical and
Electronics Engineers, Inc. Headquarters address: IEEE, 3
Park Avenue, 17th Floor, New York, NY 10016-5997, USA;
tel:
+1-212-705-8900;
http://www.comsoc.
org/ci. Responsibility for the contents rests upon authors of
signed articles and not the IEEE or its members. Unless otherwise specified, the IEEE neither endorses nor sanctions any
positions or actions espoused in IEEE Communications
Magazine.
ANNUAL SUBSCRIPTION: $27 per year print subscription. $16 per year digital subscription. Non-member print
subscription: $400. Single copy price is $25.
EDITORIAL CORRESPONDENCE: Address to: Editorin-Chief, Nim K. Cheung, Telcordia Tech., Inc., One
Telcordia Drive, Room RRC-1B321, Piscataway, NJ 088544157; tel: +(732) 699-5252, e-mail: n.cheung@ieee.org
_________.
COPYRIGHT
AND
REPRINT
PERMISSIONS:
Abstracting is permitted with credit to the source. Libraries
are permitted to photocopy beyond the limits of U.S.
Copyright law for private use of patrons: those post-1977
articles that carry a code on the bottom of the first page provided the per copy fee indicated in the code is paid through
the Copyright Clearance Center, 222 Rosewood Drive,
Danvers, MA 01923. For other copying, reprint, or republication permission, write to Director, Publishing Services,
at IEEE Headquarters. All rights reserved. Copyright © 2010
by The Institute of Electrical and Electronics Engineers, Inc.
POSTMASTER:
Send address changes to IEEE
Communications Magazine, IEEE, 445 Hoes Lane,
Piscataway, NJ 08855-1331. GST Registration No.
125634188. Printed in USA. Periodicals postage paid at New
York, NY and at additional mailing offices. Canadian Post
International Publications Mail (Canadian Distribution)
Sales Agreement No. 40030962. Return undeliverable
Canadian addresses to: Frontier, PO Box 1051, 1031 Helena
Street, Fort Eire, ON L2A 6C7
SUBSCRIPTIONS, orders, address changes — IEEE
Service Center, 445 Hoes Lane, Piscataway, NJ
08855-1331, USA; tel: +1-732-981-0060; e-mail:
address.change@
ieee.org.
____________
ADVERTISING: Advertising is accepted at the discretion of the publisher. Address correspondence to:
Advertising Manager, IEEE Communications Magazine,
3 Park Avenue, 17th Floor, New York, NY 10016.
SUBMISSIONS: The magazine welcomes tutorial or
survey articles that span the breadth of communications. Submissions will normally be approximately 4500
words, with few mathematical formulas, accompanied by
up to six figures and/or tables, with up to 10 carefully
selected references. Electronic submissions are preferred, and should be sumitted through Manuscript
Central
(http://commag-ieee.manuscript
central.com/). Instructions can be found at: http://
______
___
www.comsoc.org/pubs/commag/sub_guidelines.html.
__________________________
For further information contact Steve Gorshe,
Associate
Editor-in-Chief
(steve_gorshe@pmc__________
sierra.com). All submissions will be peer reviewed.
_____
4
Communications
IEEE
A
BEMaGS
F
66 SAFETY ASSURANCE AND RESCUE COMMUNICATION SYSTEMS IN
HIGH-STRESS ENVIRONMENTS: A MINING CASE STUDY
Effective communication is critical to the success of response and rescue operations.
However, unreliable operation of communication systems in high-stress environments
is a significant obstacle to achieving this. The contribution of this article is threefold.
First, it outlines those common characteristics that impair communication in highstress environments and then evaluates their importance, specifically in the
underground mine environment. Second, it discusses current underground mine
communication techniques and identifies their potential problems. Third, it explores
the design of wireless sensor network based communication and location sensing
systems that could potentially address current challenges.
PRASANT MISRA, SALIL KANHERE, DIETHELM OSTRY, AND SANJAY JHA
TOPICS IN INTEGRATED CIRCUITS FOR COMMUNICATIONS
SERIES EDITORS: CHARLES CHIEN, ZHIWEI XU, AND STEPHEN MOLLOY
74 GUEST EDITORIAL
76 VIDEO ENCODER DESIGN FOR HIGH-DEFINITION 3D VIDEO COMMUNICATION
SYSTEMS
The authors present an overview of 3D video coding standards developments and
design challenges of an MVC encoder. Then the algorithm and architecture
optimization schemes are proposed. For the trade-off between system memory
bandwidth and on-chip memory size, a cache-based prediction engine is proposed
to ease both design challenges.
PEI-KUEI TSUNG, LI-FU DING, WEI-YIN CHEN, TZU-DER CHUANG, YU-HAN CHEN, PAI-HENG HSIAO,
SHAO-YI CHIEN, AND LIANG-GEE CHEN
88 AN EMBEDDED 65 NM CMOS BASEBAND IQ 48 MHZ–1 GHZ DUAL
TUNER FOR DOCSIS 3.0
The authors present an embedded CMOS digital dual tuner for DOCSIS 3.0 and
set-top box applications. The dual tuner down-converts a total of ten 6 MHz Annex B
channels or eight 8 MHz Annex A channels, for a maximum data rate of 320 Mb/s in
Annex B and 400 Mb/s in Annex A mode.
FRANCESCO GATTA, RAY GOMEZ, YOUNG SHIN, TAKAYUKI HAYASHI, HANLI ZOU, JAMES Y.C. CHANG,
LEONARD DAUPHINEE, JIANHONG XIAO, DAVE S.-H. CHANG, TAI-HONG CHIH, MASSIMO BRANDOLINI,
DONGSOO KOH, BRYAN J.-J. HUNG, TAO WU, MATTIA INTROINI, GIUSEPPE CUSMAI, ERTAN ZENCIR,
FRANK SINGOR, HANS EBERHART, LOKE TAN, BRUCE CURRIVAN, LIN HE, PETER CANGIANE,
AND PIETER VORENKAMP
98 INTEGRATED ELECTRONIC SYSTEM DESIGN FOR AN IMPLANTABLE WIRELESS
BATTERYLESS BLOOD PRESSURE SENSING MICROSYSTEM
A wireless, batteryless, less invasive blood pressure sensing microsystem based on
an instrumented circular cuff has been developed for advanced biological research.
The proposed sensing technique avoids vessel penetration and substantially
minimizes vessel restriction due to the soft cuff elasticity.
PENG CONG, WEN H. KO, AND DARRIN J. YOUNG
106 POWER LINE COMMUNICATION NETWORKS FOR LARGE-SCALE CONTROL AND
AUTOMATION SYSTEMS
Power line communications uses the existing power line infrastructure for
communication purposes. While the majority of recent contributions have discussed
PLC for high-data-rate applications like Internet access or multimedia communication
serving a relatively small number of users, in this article the authors are concerned
with PLC as an enabler for sensing, control, and automation in large systems
comprising tens or even hundreds of components spread over relatively wide areas.
GERD BUMILLER, LUTZ LAMPE, AND HALID HRASNICA
114 IMS-COMPLIANT MANAGEMENT OF VERTICAL HANDOFFS FOR MOBILE
MULTIMEDIA SESSION CONTINUITY
The authors propose an original solution for session continuity based on the primary
design guideline of cleanly and effectively separating the signaling plane (for session
reconfiguration via SIP) from the media delivery plane (data transmission and related
handoff management operations). Our optimized handoff management techniques
exploit terminal-based decentralized predictions to minimize service-level handoff delays.
PAOLO BELLAVISTA, ANTONIO CORRADI, AND LUCA FOSCHINI
122 REPUTATION ESTIMATION AND QUERY IN PEER-TO-PEER NETWORKS
Many peer-to-peer systems assume that peers are cooperative to share and relay data.
But in the open environment of the Internet, there may be uncooperative malicious
peers. To detect malicious peers or reward well behaved ones, a reputation system is
often used. The authors give an overview of P2P reputation systems and investigate two
fundamental issues in the design: reputation estimation and query. They classify the
state-of-the-art approaches into several categories and study representative examples
in each category.
XING JIN AND S.-H. GARY CHAN
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
IEEE
IEEE
BEMaGS
F
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
THE PRESIDENT’S PAGE
TECHNICAL ACTIVITIES: STIMULATING TECHNICAL CONTENT CREATION
T
echnical Activities is the foundation
pation. Long-time members might already be
for many of the Communications Socifamiliar with many of the products, services,
ety’s products and services, while also proand opportunities offered by ComSoc. This is
viding “community” and “networking”
not so for the new members whom we need
opportunities for its participants. ComSoc’s
to attract and involve so that we can continue
Technical Activities community establishes
to grow and thrive in the coming years.
an environment for technical content creThrough various means, we plan to publicize
ation, whereas Publications and Conferand make known “open call” opportunities
ences deal with the presentation and
for all individuals to actively participate and
conversion of technical content into techniserve in ComSoc. Also, recognizing the
cal products and services. Within ComSoc,
changing and evolving technological landthe term “technical activities” encompasses
scape, we will strive to stay relevant and proa wide range of topics: technical commitvide more technical activities of interest to
BYEONG GI LEE
tees, emerging technologies, awards, educapotential members from industry.
tion, standards, distinguished lecturers,
TECHNICAL COMMITTEES
evaluation of IEEE Fellow nominations,
We currently have 25 Technical Commitand communications history. These areas
tees (TCs), including comparatively young
continue to evolve in spite of and in the face
TCs such as those on Cognitive Networks
of cuts in important parts of the ComSoc
and Power-Line Communications, and the
budget (due to the economic downturn). As
most recent addition, e-Health, created in
mentioned in the January 2010 message, we
2009. TCs help define and implement the
will strive to maintain and improve upon the
technical directions of the Society. They help
technical quality of ComSoc products and
organize and ensure the technical quality of
services, providing value to our members,
workshops and conference sessions, evaluate
benefiting the broader communications and
and endorse proposals for new events, help
networking community, and ultimately doing
implement publications, define and promote
our part in achieving ComSoc’s goal of
industry standards, and provide many pro“Serving Humanity” – for example, helping
fessional networking opportunities.
serve fundamental human needs for content
A list of ComSoc TCs and their officers
and communications.
MARK KAROL
is at http://ww2.comsoc.org/about/commitSharing this month’s column is Mark
tees/ Technical. You can learn much more
Karol, ComSoc’s Vice President — Techni___
about the TCs by following the links to their respective TC
cal Activities. Mark received his Ph.D. in electrical engiwebsites. Please visit the websites of the TCs of your interneering from Princeton University in 1986 and is an IEEE
est, contact the officers with your questions, and volunteer
Fellow. From 1985 until 2000 he was a member of the
to help them in their endeavors. Participation in one or
Research Communications Sciences Division at Bell Labomore TCs will move you into an exciting technical life,
ratories. From 2000 until 2008, he was a research scientist
where you can enjoy networking and co-working with colwith Avaya Labs. Since 2008, he has been a senior scientist
leagues having common technical interests.
in Applied Research at Telcordia Technologies. Mark
So that the TCs stay current in technical topics and
received the Society’s Donald W. McLellan Meritorious
content, we have mechanisms to control the entry and exit
Service Award in 2005, and has served as Associate Editor
of TCs, with the former performed by the Emerging Techfor the Journal of Lightwave Technology, General Chair of
nologies Committee (see the next section) and the latter
two major IEEE conferences, ICC ’02 and INFOCOM ’94,
by the TC Recertification Committee (TCRC). To ensure
ComSoc’s Chief Information Officer (CIO), Director of
that TCs maintain high quality standards and stay responMagazines, and Vice President — Conferences. In 2006sive to the interests of our technical community, the
2007, Mark was elected by ComSoc members to serve on
TCRC, chaired by VP-Technical Activities, Mark, evaluthe IEEE Board as Division III Director, representing
ates each TC once every three years and recommends to
Communications Technology.
the Board of Governors whether to re-certify or terminate
In the rest of this article, Mark and I provide a brief
it. For the recertification, the TCRC reviews the overall
overview of the many technical activities within ComSoc and
TC structure and evaluates the accomplishments, visions,
the distinguished individuals that lead the efforts. In all of
and goals of each TC. The TCRC may also recommend
these activities, there are opportunities in 2010–2011 to
that the Board of Governors create new TCs if it sees a
implement the “ComSoc Golden Triangle” vision of Globalgap in current coverage of technical areas.
ization, development of Young Leaders, and Industry partici-
6
Communications
IEEE
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
THE PRESIDENT’S PAGE
EMERGING TECHNOLOGIES
We encourage and provide assistance to members with
common interest in a new technology area to form a small
activity group, called a sub-committee, with the expectation that such a group may eventually evolve into a technical committee. The Emerging Technologies Committee,
chaired by Naohisa Ohta, helps identify and nurture new
technical directions, and approves the formation of subcommittees. The goal is timely dissemination of technical
information in emerging technology areas that are of high
interest to our members and others in our field.
Current technical subcommittees are: (i) Applications of
Nanotechnologies in Communications; (ii) Autonomic Communications; (iii) Consumer Networking; (iv) Human Centric
Communications; (v) Integrated Fiber & Wireless Technologies; (vi) Nano-Scale, Molecular, & Quantum Networking;
and (vii) Peer-to-Peer Networking. More information about
the technical subcommittees is at http://ww2.comsoc.org/about/
committees/emerging. Discussion is also under way regarding
______________
new opportunities in other new technological areas, such as
smart grids (for the power industry), vehicular networking,
and intelligent transportation systems. You may write a proposal to the ETC Chair if you identify a new emerging area
that should be of interest to ComSoc.
AWARDS
Awards, as well as Fellow evaluation and Distinguished
Lecturer selection, are auxiliary technical functions that
establish an environment for technical activities by recognizing distinguished achievements. The Awards Committee
receives nominations and evaluates candidates for ComSoc-level awards to select recipients. Vince Poor is Chair
of the Awards Committee. There are four Career Awards,
four Service Awards, and nine Paper Awards. Career
Awards recognize career-long achievements, Service
Awards recognize distinguished services to ComSoc, and
Paper Awards recognize the best papers published in various ComSoc journals and magazines.
A complete list of the ComSoc-level awards is at
http://ww2.comsoc.org/about/memberprograms/comsocawards. Please look at the list and consider nominating
_____
worthy candidates for the awards. The process (including
deadlines) for nominating candidates is included in a link
on the ComSoc website. Award winners are typically
announced during ceremonies at our annual ICC and
GLOBECOM conferences. We try to widely publicize the
nomination deadlines so that appropriate candidates are
brought to the attention of the Awards Committee. If you
know someone worthy of an award, please take the initiative and nominate them — don’t assume someone else will.
EDUCATION
ComSoc’s Education activities apply our technical products and services to educational purposes. The ComSoc
Education Board is responsible for establishing policies
and setting strategic directions for all continuing education
products and services. It is also responsible for maintaining
the vitality and quality of existing programs, assuring that
such programs continue to meet the current needs of
members (and others), and coordinating ComSoc efforts
with other IEEE education programs. Stefano Bregni
serves as Director of Education.
Examples of ComSoc educational programs include online
webinars and tutorials (e.g., the Tutorials NowTM program
features more than 90 titles), conventional tutorials at conferences, and courses especially tailored to cover certification
program topics (e.g., in Wireless Communications Engineering Technologies (WCET)). Working groups on the Education Board are working on specific tasks to monitor and
enhance all ComSoc educational products and services; for
example, exploring ways of virtual collaboration for e-teaching.
Globalization is a major value. Online tutorials and OnLine Distinguished Lectures are an excellent example of ComSoc’s efforts to facilitate participation by student chapters and
from disadvantaged areas. We will continue to search for new
educational opportunities, including development of products
aimed at students — our future Young Leaders.
STANDARDS
ComSoc sponsors standards activities to meet the needs
of our members and the worldwide communications industry. The main goal of ComSoc’s standards activities is the
development of technically excellent and widely adopted
standards in communications, networking, and related
fields. Standards projects of interest to ComSoc include
issues related to power-line communications, broadband
over power-line networks, smart grid technologies, Ethernet passive optical networks, and fundamental aspects of
telephony. ComSoc provides procedural guidelines, discussion forums, and support services for approved standards
activities. A detailed description of standards activities is
available at http://ww2. comsoc.org/about/documents/
pp/5.14.
_____
ComSoc standards activities are conducted at the initiative of members and/or organizations within the Society, or
of individuals or entities outside of the society. The activities
are led by the Director of Standards, Curtis Siller, who
helps influence and guide standards development. The standards development activity is conducted in partnership with
the IEEE Standards Activities organization, with appropriate liaison with and contributions to other standards bodies.
By their very nature, standards activities engage industry
participation in ComSoc. In consideration of the importance
of standards as an ultimate product of technical findings and
also the significant influence of standards on industry, we
are carefully studying the possibility of upgrading ComSoc’s
standards activities to Vice President level.
SELECTION OF DISTINGUISHED LECTURERS
ComSoc selects Distinguished Lecturers who will provide lectures to audiences arranged by ComSoc Chapters
all over the world. The Distinguished Lecturers Program
(DLP) is one of a large set of membership programs, developed for the benefit of the members (and others in our
field). Distinguished Lecturer Tours, especially when conducted in developing countries, help implement ComSoc’s
vision of globalization and fosters ComSoc’s mission of presentation and exchange of information among its members
and technical communities throughout the world. A DLT
can often help generate ComSoc membership and the for-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
7
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
THE PRESIDENT’S PAGE
mation of new Chapters. Selection of Distinguished Lecturers is performed by the Distinguished Lecturers Selection
Committee, which is chaired by the Vice-Chair and Secretary of the Technical Activities Council, Mehmet Ulema.
As in other programs of ComSoc, we have an “open
call” nomination process for Distinguished Lecturers; information is available at http://ww2.comsoc.org/ about/documents/pp/6.8. Currently ComSoc has 19 Distinguished
Lecturers (DLs), as listed at http://ww2.comsoc.
org/about/memberprograms/distinguished-lecturers. There
also is a link to a list of past Distinguished Lecturers and
their topics, who are encouraged to continue giving lectures
(after their two-year terms have expired) in conjunction
with business trips or events that happen to bring them to
ComSoc Chapters of interest. We also have recorded a
small, but hopefully growing, number of On-Line Distinguished Lectures (see the above-referenced website).
neer/scientist, or technical leader. The total number of new
IEEE Fellows in any one year is at most one-tenth of one
percent of the IEEE voting membership (i.e., several hundred new Fellows per year). Further information (including qualifications and deadlines) about the IEEE Fellow
program is available at http://www.ieee.org/fellows.
A key part of the selection process is the evaluation of
candidates by each of the IEEE’s Technical Societies. In
ComSoc, the Fellow Evaluation Committee, chaired by
Russell Hsing, reviews and evaluates all candidates identified as working in our fields of interest. The evaluations,
including a rank order of the candidates and a summary of
the evaluation of each candidate, are then forwarded to
the IEEE Fellow Evaluation Committee. For further information about the Fellow Evaluation Committee, refer to
http://ww2.comsoc.org/about/
documents/pp/6.2.
________________________________
EVALUATION OF IEEE FELLOW NOMINATIONS
Communications History is a special auxiliary function of
technical activities. It deals with the history of communications technology, recognizing milestones of technological
development. This function is conducted by the Communications History Committee, chaired by Mischa Schwartz. This
committee is responsible for identifying, placing in electronic
archives, and raising public awareness through all appropriate
steps on the most important facts/persons/achievements of
communications history in particular, as
well as telecommunication milestones in
general. One important activity of this
committee is the regular publication of
“History of Communications” articles in
IEEE Communications Magazine. In
addition, the committee has organized
special sessions on “communications
history” at recent IEEE GLOBECOM
conferences. More are planned for the
future. Please contact Mischa if you
happen to find any important findings
in communications history that have not
been duly recognized.
IEEE Fellow status is granted to a person with an
extraordinary record of accomplishments in any of the
IEEE’s designated fields of interest. The honor is conferred by the IEEE Board of Directors. It is one of the
highest honors that can be bestowed upon an individual by
the Institute. It recognizes important contributions as an
application engineer/practitioner, educator, research engi-
__________
8
Communications
IEEE
COMMUNICATIONS HISTORY
__________
In summary, technical activities covers a wide range of programs and helps
form the backbone of the IEEE Communications Society. We are fortunate
to have an experienced and dedicated
group of individuals who help lead and
guide the various activities. However,
we always welcome new volunteers to
join us and make our programs even
better. In keeping with ComSoc’s Golden Triangle vision, we especially encourage additional participation by young
members and industrial participants —
from all parts of the world. A good first
step is to join a Technical Committee
(which is easy to do and has no membership fees); many of the current ComSoc leaders got their start by
volunteering to help in ComSoc TCs.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
_____________________________________________________
until June 30 2010
Presented at IEEE GLOBECOM 2009 by Ali Akansu
Discrete Fourier Transform (DFT) has been the center piece of popular technologies spanning xDSL based
high-speed Internet access to OFDM based wireline and wireless communications. This tutorial reviews the
recently introduced non linear phase Generalized DFT, which offers significant improvements over linear phase
DFT. Examples include Adaptive GDFT's ability to mitigate degradation of RF power amplifier performance and
Bit Error Rate Performance. Several efficient GDFT design methods and their performance in real-world communications scenarios are highlighted.
FREE ACCESS
SPONSORED BY
For other sponsor opportunities, please contact Eric Levine, Associate Publisher Phone: 212-705-8920, E-mail: ______________
e.levine@comsoc.org
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SOCIETY NEWS
SOCIETY MEMBERS NAMED TO FELLOW GRADE
Election to the grade of IEEE Fellow is one of the highest honors that can be
bestowed upon our members by the Institute in recognition of their technical, educational,
and leadership achievements. Only a select few IEEE members earn this prestigious
honor.
Congratulations to the following Communications Society members for their election to the grade of Fellow of the IEEE. They now join company with a truly distinguished roster of colleagues.
RAJ ACHARYA
For contributions to
biomedical imaging
and bioinformatics.
EITAN ALTMAN
For contributions
to analysis,
optimization, and
control of
telecommunication
networks.
JOSEPH BERTHOLD
For leadership
in optical
internetworking.
EDGAR CALLAWAY
For contributions to
wireless sensor
networks and low
power design
techniques for
communications
devices and systems.
HSIAO-HWA CHEN
For contributions to
radio resource
allocation in code
division multiple
wireless systems.
10
Communications
IEEE
ZHIZHANG (DAVID) CHEN
For contributions to
time-domain
electromagnetic
modeling and
simulation.
ROY CIDECIYAN
For contributions to
signal processing and
constrained coding
for magnetic
recording.
THOMAS CLOONAN
For leadership in
development of cable
modem termination
systems.
ROBERT DOVERSPIKE
For contributions
to architectures,
modeling, and
optimization of
telecommunication
networks.
HESHAM EL-GAMAL
For contributions
to multiple-input
multiple-output and
cooperative
communications.
MARC GOLDBURG
For leadership in the
development and
commercialization
of spectrally efficient
wireless
communications
systems.
JOSEPH HELLERSTEIN
For contributions to
control engineering
for performance
management of
computing systems.
SHIVKUMAR KALYANARAMAN
For contributions to
traffic management
in computer
communication
networks.
POOI-YUEN KAM
For contributions to
receiver design and
performance analysis
for wireless
communications.
YOSHIO KARASAWA
For contributions
to the measurement
and modeling of
propagation effects in
radio communication
systems.
GERHARD KRAMER
For contributions to
coded modulation,
iterative decoding,
and cooperative
communications.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
SOCIETY NEWS
UWE KRAUS
For contributions to
digital television
signal compression
and transmission.
MARWAN KRUNZ
For contributions
to resource
management policies
in wireless networks.
CHANG-HEE LEE
For contributions to
wavelength division
multiplexed-passive
optical network.
PING LI
For contributions to
iterative signal
processing, multi-user
detection and
concatenated error
control codes.
WANJIUN LIAO
For contributions to
communication
protocols in
multimedia
networking.
JOHN C.S. LUI
For contributions to
performance
modeling and analysis
of storage
communication
systems and peer-topeer networks.
STAN LUMISH
For leadership in the
development and
implementation of
commercial
terrestrial lightwave
systems.
TADASHI MATSUMOTO
For contributions
to signal processing
for wireless
communications.
VICTOR MILLER
For contributions
to elliptic curve
cryptography.
JOSEPH MITOLA
For contribution to
software-defined and
cognitive radio
technologies.
PRASANT MOHAPATRA
For contributions to
the quality of service
provisioning in
computer networks.
ARIA NOSRATINIA
For contributions
to multimedia
and wireless
communications.
HARRY PERROS
For contributions to
performance
evaluation modeling
of computer
networks.
CHUNMING QIAO
For contributions to
optical and wireless
network architectures
and protocols.
RAMESH RAO
For leadership
in wireless
communications.
A. L. NARASIMHA REDDY
For contributions to
multimedia storage
and network support.
LESLIE RUSCH
For contributions in
optical and wireless
communications
systems.
CHRISTIAN SCHLEGEL
For contributions
to iterative
demodulation and
decoding in wireless
communication.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
11
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SOCIETY NEWS
ROBERT SCHOBER
For contributions
to wireless
communications.
DAN SCHONFELD
For contributions
to image and
video analysis.
MARK SHAYMAN
For contributions to
the theory of Riccati
equations and
discrete-event
dynamic systems.
ANDREW SINGER
For contributions to
signal processing
techniques for digital
communication.
VISHWANATH SINHA
For contributions to
electrical engineering
education.
PETER STEENKISTE
For contributions to
optimization and
monitoring
techniques for
distributed
communication
systems.
12
Communications
IEEE
MILICA STOJANOVIC
For contributions to
underwater acoustic
communications.
HEINRICH STUTTGEN
For leadership in
industrial research.
WEI SU
For leadership
in military
communications and
electronic warfare
technologys.
EMMANOUIL M. TENTZERIS
For contributions to
three dimensional
conformal integrated
devices for wireless
communications
and sensing.
NIAN-FENG TZENG
For contributions to
parallel computer
systems and
scalable routers.
NITIN VAIDYA
For contributions to
wireless networking
protocols and mobile
communications.
CHENGSHAN XIAO
For contributions to
channel modeling
and signal processing
for wireless
communications.
HOWARD YANG
For leadersip in
mixed-signal
integrated circuit
design and
manufacturing.
FENG ZHAO
For contributions to
networked embedded
computing and
sensor networks.
WENWU ZHU
For contributions
to video
communications
over the internet
and wireless.
ZORAN ZVONAR
For leadership in the
development of
digital signal
processing software
and hardware for
wireless cellular
communication.
MAHESH VARANASI
For contributions to multi-user and
wireless communication theory.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
IEEE
IEEE
BEMaGS
F
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
CONFERENCE CALENDAR
2010
APRIL
• ICT 2010 – 17th Int’l. Conference on
Telecommunications, 4–7 April
Doha, Qatar. Info: http://www.qu.edu.qa/
ICT2010/index.php
____________
z IEEE DYSPAN 2010 – IEEE Int’l.
Symposium on Dynamic Spectrum
Access Networks, 6–9 April
Singapore. Info: http://www.ieee-dyspan.org
• WAMICON 2010 – IEEE Wireless and
Microwave Technology Conference,
12–13 April
Dallas, TX. Info:________________
http://www.ewi.info/world-
Faro-Algarve, Portugal. Info: _________
http://noc-con-
wide-cybersecurity-summit
________________
ference.com/
________
z IEEE LAN/MAN 2010 - 17the IEEE
Workshop on Local and Metropolitan Area Networks, 5-7 May
• HPSR 2010 - 2010 Int’l. Conference
on High Performance Switching and
Routing, 14-16 June
Long Branch, NJ. Info: ___________
http://www.ieeelanman.org
_______
Dallas, TX. Info: http://opnear.utdallas/edu/
activ/hpsr2010/index.html
________________
• ISWPC - Int’. Symposium on Wireless Pervasive Computing, 5-7 May
• APSITT 2010 - 8th Asia-Pacific Symposium on Information and Telecommunication Technologies, 15-18 June
Modena, Italy. Info: http://www.iswpc.org/
2010/
___
z IEEE CTW 2010 - 2010 IEEE Communication Theory Workshop, 1012 May
Cancun, Mexico. Info: http://www.ieee-
Melbourne, FL. Info: ____________
http://www.wamicon.org/
_____
c
t
w
.
__________________________
• IEEE SARNOFF 2010 - 23rd IEEE
SARNOFF Symposium 2010, 12-14
April
• CNSR 2010 - Communication Networks and Services Research 2010
Conference, 12-14 May
Princeton, NJ. Info: http://ewh.ieee.org/r1/
princeton-centraljersey/2010_Sarnoff_Sym__________________________
posium/
_____
Montreal, Canada. Info: http://cnsr.info/
cnsr2010/
______
org/
__
• EW 2010 – European Wireless 2010,
12–15 April
z IEEE ICC 2010 - IEEE Int’l.
Conference on Communications,
23-27 May
Lucca, Italy. Info: http://www.european______________
wireless2010.org
__________
Capetown,
South
Africa.
http://www.ieee-icc.org/2010/
• IEEE RFID 2010 - IEEE Int’l. Conference on RFID 2010, 14-15 April
JUNE
Orlando, FL. Info: http://www.ieee-rfid.org/
2010
___
z IEEE WCNC 2010 - IEEE Wireless
Communications and Networking
Conference, 18-21 April
Sydney, Australia. Info: __________
http://www.ieee______
wcnc.org/
z IEEE/IFIP NOMS 2010 – IEEE/IFIP
Network Operations and Management Symposium, 19–23 April
Osaka, Japan.
• WTS 2010 - Wireless Telecommunications Symposium 2010, 21-23 April
Tampa, FL. Info: http://www.csupomona.edu/
wtsi
___
MAY
• WCS 2010 - First Worldwide
Cybersecurity Summit, 4-5 May
Info:
• NGI 2010 - 6th Euro-NF Conference
on Next Generation Internet, 2-4 June
Paris, France, Info: http://euronf.enst.fr/
NGI2010/Home.html
_____________
• CTTE 2010 - 9th Conference of
Telecommunication, Media and Internet Techno-Economics, 7-9 May
Ghent, Belgium. Info: http://www.ctte-con____________
ference.org/
_______
z IEEE CQR 2010 - 2010 Int’l. Communications Quality and Reliability
Workshop, 8-10 June
Vancouver,
BC,
Canada.
http://www.ieee-cqr.org/
Info:
• NOC/OC&I 2010 - 15th European
Conference on Networks and Optical
Communications & 5th Conference on
Optical Cabling and Infrastructure, 810 June
z Communications Society portfolio events are indicated with a diamond before the listing;
• Communications Society technically co-sponsored conferences are indicated with a bullet
before the listing. Individuals with information about upcoming conferences, calls for papers, meeting
announcements, and meeting reports should send this information to: IEEE Communications
Society, 3 Park Avenue, 17th Floor, New York, NY 10016; e-mail: b.erlikh@comsoc.org;
____________ fax:
+1-212-705-8999. Items submitted for publication will be included on a space-available basis.
14
Communications
IEEE
Kuching, Malaysia. Info: http://www.ieice.org/
~in_ac/APSITT/2010
_____________
z IEEE IWQOS 2010 - 18th IEEE Int’l.
Workshop on Quality of Service, 1618 June
Beijing, China. Info: ___________
http://www.ieeeiwqos.org/
______
• ICUFN 2010 - 2nd Int’l. Conference
on Ubiquitous Networks and Future
Networks, 16-18 June
Juju Island, Korea. Info: http://www.icufn.org/
• IEEE ISGT 2010 - 2010 IEEE Innovative Smart Grid Technologies Conference, 19-21 June
Gaithersburg, MD.
z IEEE SECON 2010 - 2010 IEEE
Communications Society Conference on Sensor, Mesh and Ad Hoc
Communications and Networks, 2125 June
Boston, MA. Info: http://www.ieee-secon.org/
2010/index.html
__________
• Med-Hoc-Net 2010 - 9th IFIP Annual
Mediterranean Ad Hoc Networking
Workshop, 21-25 June
Juan Les Pins, France. Info: __________
http://www.medhoc-net-2010.org/
___________
z ISCC 2010 - IEEE Symposium on
Computers and Communications,
22-25 June
Riccione, Italy. Info: http://www.ieee-iscc.org/
2010
___
JULY
z IEEE HEALTHCOM 2010 - IEEE
12th Int’l. Conference on e-Health
Networking, Application & Services,
1-3 July
Lyon, France. Info: ______________
http://www.ieee-healthcom.org/
_____
• SPECTS 2010 - 2010 Int’l. Symposium on Performance Evaluation of
Computer and Telecommunication
Systems, 11-14 July
Ottawa, Canada. Info: http://atc.udg.edu/
SPECTS2010
________
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
_____________________________________________________
ADVANCED OPTICAL FORMATS
OF 40/100G AND BEYOND
Coherent detection and advanced signal processing are
required for the test and measurement of advanced
modulation formats that are currently realized for optical
transmission of 40/100G and beyond. In this presentation the
basic hardware building blocks and the necessary algorithms
are introduced. New tools to characterize and quantify the
signal quality are applied to a number of exemplary signals.
FREE ACCESS
FOR A LIMITED TIME ONLY
UNTIL MAY 31
FREE ACCESS
SPONSORED BY
For other sponsor opportunities, please contact Eric Levine, Associate Publisher Phone: 212-705-8920, E-mail: ______________
e.levine@comsoc.org
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
NEW PRODUCTS
AGILENT TECHNOLOGIES’ NEW LTE
APPLICATIONS TARGET 4G
SYSTEM-LEVEL DESIGNERS
Agilent Technologies
Agilent Technologies Inc. has introduced a new line of system-level design
and verification products for 3GPP
LTE physical layer (PHY) design. In
addition to its traditional test and measurement products, Agilent now provides predictive products and
algorithmic references for the SystemVue platform that are consistent
with the LTE v.8.9.0 (December 2009)
standard. The new line includes four
products that accelerate 4G deployment
for LTE system architects, baseband
hardware designers, and RF equipment
by bringing new levels of realism into
the architecture and modeling stages.
While 4G networks promise dramatic improvements in data throughput
and spectral efficiency, the complexity
of the evolving 3GPP LTE standard has
forced many system architects to reconsider their use of general-purpose
toolsets. By cutting across a variety of
domains (such as signal processing and
mathematics, baseband hardware
design, RF/analog design and measurements), Agilent's new 4G products are
able to provide system architects with
the focused application support for
LTE v.8.9.0 that they demand. Agilent's
4G products streamline verification and
bring measurement-level realism into
the creative process. The result is higher-performance Layer 1 architectures
and algorithms resulting in earlier
design maturity and the need for less
design margin.
The four new products being introduced by Agilent include:
The W1715 MIMO Channel Builder
is a simulation blockset for LTE architecture and receiver designers, based on
the WINNER and WINNER-II fading
algorithms. By incorporating non-ideal
MIMO antenna performance (e.g.,
crosstalk and directionality), the W1715
goes beyond these standard propagation models. It enables 2D far-field
data to be imported from antenna measurements or 3D EM simulations,
including Agilent EM PRO.
The W1716 Digital Pre-Distortion
builder helps LTE system integrators,
RF component designers, and baseband
architects quickly transition from 3G to
4G by creating baseband signal processing networks that improve the range of
analog power amplifiers and transceiver
ICs, improve efficiency, and extend battery life. The W1716 also quickly assesses the suitability of existing 3G designs
for 4G applications. Such capabilities
16
Communications
IEEE
translate into savings for both design
effort and component cost.
The W1910 LTE Baseband Verification Library reference blockset has
been updated to LTE v.8.9.0 and now
includes expanded PRACH and HARQ
support. The HARQ simulation support uses a unique dynamic dataflow
simulation mode. This mode allows the
symbol rate to change dynamically over
the course of the simulation while
retaining the timing and carrier information necessary for full RF effects,
frequency-dependent phase noise, and
channel fading.
www.agilent.com
handsets, basestations, wireless PC
cards, and a host of other embedded
solutions.
The portfolio includes the industry's
first LTE front-end modules for Bands
I, IV, VII and VIII (the SKY77445,
SKY77455, SKY77456 and SKY77458)
for worldwide applications, and the
industry's first power amplifier modules
supporting LTE-FDD for North America (the SKY77449 and SKY77453).
www.skyworksinc.com
SKYWORKS SUPPORTS CUSTOMERS’
LTE PLATFORMS WITH LINE OF
POWER AMPLIFIER AND FRONT-END
MODULES
The high-end R&S ZVA67 vector
network analyzer from Rohde &
Schwarz is now also available as a fourport model. This is the first network
analyzer on the market to feature four
test ports for measurements up to 67
GHz. Its high dynamic range (110 dB at
67 GHz) and output power (6 dBm at
67 GHz) give the R&S ZVA67 the flexibility and performance required for
characterizing components and modules
in the microwave and millimeter-wave
range. It allows users in research and
development to determine the S parameters of multiport devices quickly and
with high precision. As an extra benefit,
the analyzer’s four internal signal
sources reduce test system complexity
and the number of instruments
required, e.g. for measuring frequencyconverting DUTs, because no external
signal generators are needed.
The new R&S ZVA67, with its
unique architecture that includes four
ports, four integrated signal sources and
eight receivers, enables measurements
on multiport devices such as mixers,
couplers or balanced DUTs, with just
one instrument. With its high output
power of 6 dBm and wide power sweep
range of > 40 dB, the R&S ZVA67 is
able to characterize the small- and
large-signal behavior of active components.
Linear and nonlinear measurements
can be carried out using a simple test
setup. This advantage becomes apparent, for example, when measuring the S
parameters or intermodulation of mixers or amplifiers, or when measuring
the group delay and phase of up- or
downconverters.
Offering a frequency range up to 67
GHz, the four-port R&S ZVA67 covers
the band intended for the wireless
transmission of multimedia data (wireless HDMI). Additional applications
can be found in the aerospace and
defense sectors.
www.rohde-schwarz.com
Skyworks Solutions, Inc.
Skyworks Solutions has announced
the company's extensive portfolio of
long-term evolution (LTE) power
amplifier and front-end modules is now
supporting the M710 solution from STEricsson, a world leader in wireless
semiconductors and platforms.
LTE is emerging as the 4G standard
of choice worldwide for mobile broadband systems. These 4G systems are
expected to significantly boost network
throughput, improve spectral efficiency
and performance, reduce latency, simplify roaming and further drive
economies of scale. According to Global Mobile Suppliers Association (GSA),
LTE network commitments increased
100 percent in the last eight months
with 51 networks in 24 countries worldwide currently committed to LTE. All
major handset OEMs, infrastructure
suppliers, and operators worldwide are
now committed to this technology with
multiple trials underway.
Skyworks’ family of LTE products,
which includes four front-end modules
in addition to two power amplifier modules, provide the most complete and
flexible set of options for manufacturers
developing and building 4G-enabled
HIGH-END R&S ZVA67 VECTOR
NETWORK ANALYZER
Rohde & Schwarz
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
CALL FOR PAPERS
th
44 Annual Asilomar Conference on
Signals, Systems, and Computers
Asilomar Hotel and Conference Grounds
Pacific Grove, California
November 7-10, 2010
www.asilomarssc.org
Authors are invited to submit papers before June 1st, 2010, in the following areas:
A. Communications Systems: 1. Error Control Coding,
2. CDMA, 3. Modulation and Detection, 4. Performance Bounds,
5. Synchronization, 6. Ultra Wideband, 7. OFDM / Multicarrier,
8. Wireless Communications, 9. Optical Communications,
10. Cognitive SDR, 11. Adaptive Waveform Design
B. MIMO Communications and Signal Processing:
1. Space-Time Coding and Decoding, 2. Channel Estimation and
Equalization, 3. Multi-User and Multi-Access Methods,
4. Cooperative Diversity.
9. Compressive Sensing, 10. Information Theoretic Signal Processing,
11. Spectral Analysis
F. Biomedical Signal and Image Processing: 1. Medical Image
Analysis, 2. Imaging Modalities, 3. Advances in Medical Imaging,
4. Biomedical Signal Processing, 5. Biomedical Applications,
6. Bioinformatics, 7. Image Registration and Multi-modal Imaging,
8. Image Reconstruction, 9. Computer Aided Diagnosis, 10. Functional
Imaging, 11. Visualization
G. Architecture and Implementation: 1. Programmable and
C. Networks: 1. Transmission Techniques for Ad Hoc
Networks, 2. Wireless Sensor Networks, 3. Network Information
Theory, 4. Optical Networks
Reconfigurable Architectures, 2. SOC Architectures, 3. Low-power
Methods, 4. Compilers and Tools, 5. Integrated Algorithm and
Architecture Implementation, 6. Computer Arithmetic, 7. Numerical
Processing
D. Adaptive Systems and Processing: 1. Adaptive Filtering,
2. Fast Algorithms for Adaptive Filtering, 3. Frequency-Domain
and Subband Adaptive Filtering, 4. Adaptive Blind Processing
E. Array Processing and Statistical Signal Processing: 1.
Array Processing and Beamforming, 2. Sonar and Acoustical
Array Processing, 3. Radar Array Processing, 4. Remote Sensing,
5. Signal Separation, 6. Estimation and Detection, 7. NonGaussian and Nonlinear Methods, 8. Identification,
H. Speech, Image and Video Processing: 1. Speech Processing,
2. Speech Coding, 3. Speech Recognition, 4. Narrowband / Wideband
Speech and Audio Coding, 5. Document Processing, 6. Models for Signal
and Image Processing, 7. Image and Video Coding, 8. Image and Video
Segmentation, 9. Image and Video Analysis, 10. Image / Video Security,
Retrieval and Watermarking, 11. Image and Video Enhancement /
Filtering, 12. Biometrics and Security, 13. Wavelets
Submissions should include a 50 to 100 word abstract and an extended summary (500 to 1000 words, plus figures). Submissions must
include the title of the paper, each author's name and affiliation, and the technical area(s) in which the paper falls with number(s) from the
above list. Check the conference website (www.asilomarssc.org) for specific information on the electronic submission process. Submissions
will be accepted starting February 1, 2010. No more than FOUR submissions are allowed per contributor, as author or co-author. All
submissions must be received by June 1st, 2010. Notifications of acceptance will be mailed by mid August 2010, and author information
will be available on the conference website by late August 2010. Full papers will be due shortly after the conference and published in early
2011. All technical questions should be directed to the Technical Program Chair, Dr. Miloš Doroslovaþki, e-mail doroslov@gwu.edu, or
the General Chair, Dr. Linda DeBrunner, e-mail linda.debrunner@fsu.edu.
CONFERENCE COMMITTEE
General Chair:
Technical Program Chair:
Conference Coordinator:
Publication Chair:
Publicity Chair:
Finance Chair:
Linda DeBrunner, Florida State University
Miloš Doroslovaþki, The George Washington University
Monique P. Fargues, Naval Postgraduate School
Michael Matthews, ATK Mission Research
Murali Tummala, Naval Postgraduate School
Frank Kragh, Naval Postgraduate School
The site for the 2010 Conference is at the Asilomar Conference Grounds, in Pacific Grove, CA. The grounds border the Pacific Ocean and
are close to Monterey, Carmel, and the scenic Seventeen Mile Drive in Pebble Beach.
The Conference is organized in cooperation with the Naval Postgraduate School, Monterey, CA, and ATK Mission Research, Monterey,
CA. The IEEE Signal Processing Society is a technical co-sponsor of the conference.
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
_____________________
_________________________________________
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Global
Newsletter
April 2010
Activities of the IEEE ComSoc Nanjing Chapter
By Nan Liu, Yueming Cai, and Guangguo Bi, ComSoc Nanjing Chapter, China
The IEEE ComSoc Nanjing Chapter was founded on April
13, 2008. The opening ceremony was held in Nanjing, China.
Over 35 Chapter members and over 40 guests attended the
ceremony. After the inauguration, two keynote addresses,
“Development of IEEE in China” and “New Rising Wireless
Communications and Wireless Networks,” were given. In the
afternoon the Executive Committee of the Chapter was
founded and discussed activities in 2008. The Chapter Officers
are Prof. Guangguo Bi of Southeast University as Chapter
Chair; Profs. Jinlong Wang, Baoyu Zheng, Jinkang Zhu, and
Aiping Huang from various universities serve as Chapter
Vice-Chairs; Prof. Yueming Cai is Secretary; and Prof. Nan
Liu serves as Secretary and Treasurer.
Since coming into existence, the ComSoc Nanjing Chapter
has been very active in hosting and sponsoring various activities to serve its members. In 2008 we technically sponsored/cosponsored three conferences/workshops: the International
Conference on Communications and Networking in China
(CHINACOM 2008, co-sponsored) was held in Hangzhou in
August, accepting 270 papers for an acceptance rate of 32.8
percent, with 121 IEEE members attending the conference;
the Workshop on Cognitive Wireless Networks (sponsored),
held in Nanjing in September, with around 140 attendees,
more than 30 being Chapter members; and the Conference on
Communications and Signal Processing in China (CCSPC
2008, co-sponsored) held in October in Zhengzhou with
around 150 attendees.
In 2009 the ComSoc Nanjing Chapter was even more
active. In June 2009 the (technically co-sponsored) Conference on Green Wireless Technology and Systems was held in
Huangshan, China, with 115 attendees from many Chinese
universities and industrial companies. The main topics included power-efficient and energy-saving wireless communications
system design and engineering techniques, wireless access protocols, services, software design, and wireless resource saving.
In August 2009 the (co-sponsored) 9th Asian Conference
on Quantum Information Science (AQIS ’09) was held in
Nanjing with around 180 attendees. AQIS ’09 focused on
quantum information science and technology, bridging quantum physics, computer science, mathematics, and information
technologies.
In September 2009 we hosted three excellent talks by
IEEE ComSoc Distinguished Lecturers. The first talk,
“Resource Management for Multitier Wireless Networks,”
given by Prof. Xueming (Sherman) Shen from the University
of Waterloo, Canada, discussed issues on resource and mobility management algorithms, and achieving flexible and effective utilization of network resources with guarantees for
end-to-end QoS requirements of multimedia traffic. The sec-
The opening ceremony of WCSP 2009 in Nanjing, China.
ond talk, “Resource Allocation for Cellular/WLAN Integrated
Networks” by Prof. Weihua Zhuang, also from the University
of Waterloo, Canada, focused on ways to enhance multiservice provisioning by taking advantage of the complementary
cellular/WLAN integrated network strengths, and details of
call assignment/reassignment strategies, admission control
policies, and the impact of user mobility and data traffic variability. Both lectures drew a great deal of interest, and more
than 170 faculty members and students with more than 30
IEEE members in attendance. The third talk, given by Prof.
Jacob Gavan, titled “Concepts, Applications and Design of
High Altitude Platform Radio Relays,” expounded the feasibility, recent status, technical issues, and advantages of different categories of HAPS for several applications. The talk
concluded with recent trends in HAPS design and developments. Around 50 faculty members and students, among
whom over 10 were IEEE members, attended the lecture and
benefited from learning about HAPS.
The highlight of the activities of the ComSoc Nanjing
Chapter in 2009 was technically sponsoring the 2009 International Conference on Wireless Communications and Signal
Processing (WCSP 2009) held in Nanjing, China, on November 13–15, which attracted much attention from both the academic and industrial fields. It received a total of 1023
submissions from 32 countries and regions. All papers were
rigorously and independently peer reviewed by the 104 TPC
members and a large number of reviewers. The conference
finally accepted 366 high-quality papers from 23 countries and
regions, representing an acceptance ratio of 36 percent.
Accepted papers were grouped into 24 sessions with two
tracks, Wireless Communications and Signal Processing. A
highlight of the conference was the seven keynote speeches
delivered by distinguished scientific experts from all around
the world. About 300 researchers, scientists, engineers, and
(Continued on Newsletter page 4)
Global Communications Newsletter • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
1
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
White Spaces: Unused TV Band Channels for
Rural Broadband in the United States
By Ana Vazquez Alejos, Felipe Gil Castiñeira, Manuel Garcia Sanchez, and Iñigo Cuiñas
University of Vigo, Spain; New Mexico State University, USA
During the next few years there are three technologies
called to change together the traditional rigid policy for spectrum allocation: software defined radio (SDR), cognitive radio
(CR), and white space devices (WDs). The CR paradigm proposes the creation of intelligent devices (usually developed
with SDR platforms so that they can modify the radio parameters by software) which automatically change their conditions
of operation in response to user demands or changes in the
surrounding network. The application of these technologies in
the channels reserved as guard bands, known as white spaces,
is derived in WDs, which have made possible not only the use
of the unused bands, but also the use of licensed bands by
unlicensed users without causing interference to the legitimate
owners.
However, in some cases spectrum-sensing technology may
not be completely effective in preventing co- or adjacentchannel interference with the licensed user. In order to solve
this problem, WDs have been coupled with geolocation technology (e.g., GPS positioning along with a database of services
and operation times). In November 2009 the U.S. Federal
Communications Commission (FCC) launched a call for proposals to manage a database network that registers and controls the use of an important amount of liberated spectrum
allocated in the 50–3000 MHz band to prevent the emerging
new WDs from interfering with TV stations and other wireless
services operating in this frequency region.
But is there available space in so large and attractive a
region of the extremely scarce frequency spectrum? The reality is that a non-negligible number of gaps exist between
50–3000 MHz since the analog TV switchoff process started,
and this opportunity did not pass unappreciated by many eyes
(Fig. 1). The Wireless Innovation Alliance [1], formed by
companies such as Google, Dell, Hewlett-Packard, Motorola,
and Microsoft, became the impelling motor for the free use of
this accessible spectrum to develop a new generation of wireless devices that could be the answer to the need for providing
broadband access to U.S. rural areas in a cost-effective manner. The first pilot was run in October 2009 in Claudville, Virginia [2].
There are several standardization organizations working on
white space technology. The IEEE [3] formed a work group
that is closing the P1900 standard, which will try to provide
guarantees for compatibility and interference avoidance. Even
the WiMAX Forum has announced a modification to accommodate its technology to the white spaces in an attempt to
avoid losing the train of a business estimated in hundreds of
billions of dollars.
The unused TV broadcast channels were liberated as a
result of the evolution toward digital terrestrial television
(DTT) presenting blocks of 6 MHz, although this assignment
varies all over the territory. The origin of the changes in the
spectrum scene began with the abandonment of channels 52
to 69, or 698–806 MHz, by broadcasters as a response to the
analog switchoff. This portion constitutes the 700 MHz spectrum publically auctioned in 2008 to 100 bidders that obtained
about 1100 licenses for $20 million. These licensed spaces will
provide mainly cellular expansion, including fourth-generation
Long Term Evolution (4G LTE), commercial mobile TV, and
wireless broadband.
The assignment of frequencies to DTT stations was done
in the space 54–698 MHz, channels 2–51, and due to the local
feature of these new assigned frequencies many gaps can be
found in this region of the spectrum, mainly due to channel
2
Communications
IEEE
Figure 1. White space channels available in December 2009 [4].
guards. Since the FCC announced the free use of part of the
unused broadcast channels in 2008, the technological community has put forth a giant effort to satisfy all the parts involved.
The main opposition derived from the TV broadcasters that
predicted important malfunctions and interference in their
licensed frequencies due to the WDs. A pilot developed in
2008 demystified the prognosticated problems of co- and adjacent-channel interference, showing that cognitive technologies
are the answer for future services and gadgets operating in
white space gaps [2].
Among the technical advantages offered by these gaps allocated under 900 MHz, despite their unlicensed status, we can
mention the longer coverage areas that can be achieved (up to
one mile), the reduced loss by penetration into buildings and
vegetation (a detail especially relevant in rural areas), and the
reduced impact of multipath, at least compared to frequencies
of the 2.4 and 3.5 GHz bands. The availability of low-cost
hardware is not less important, due to the facilities for the
semiconductor technology at ultra high frequency (UHF).
Nevertheless, not all the news is good. We cannot forget that
longer antennas are required, and this fact can represent a business opportunity for antenna manufacturers. Another downside
could be the need for incorporating geolocation in the WDs,
increasing power consumption and size.
In the future, the WDs and unlicensed services operating
in the unused broadcast frequencies will operate under the
guidelines given by CFR 47 Part 15 and amendments, retaining the 6 MHz channel. More spectrum space is available in
the upper 700 MHz segment, which is expected to be assigned
to deploy a nationwide communication network for public
safety forces. This space contains 10 MHz blocks.
The creation of the database proposed by the FCC will prevent licensed services, such as wireless microphone users or TV
broadcasters, suffering interference during their operation. The
standards under development will ensure that the new devices
have dynamic access control, taking into account the local spectrum conditions to modify transmission/reception. The database
to be implemented plays an important role.
2010 can present the definitive and successful launch of
WhiteFi (a Wi-Fi style system operating in white spaces), thus
providing full access to the information society for rural areas.
References
[1]
[2]
[3]
[4]
The Wireless Innovation Alliance
Bridge Wave Inc
www.ieee802.org/22
http://www.showmywhitespace.com
Global Communications Newsletter • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Hightlights of the IEEE 9th TELSIKS Conference
By Prof. Bratislav Milovanovic, University of Nis, Serbia
On October 7–9, 2009 the Faculty of Electronic Engineering in Nis, Serbia, hosted fir the ninth time the biennial International Conference on Telecommunications in Modern
Satellite, Cable and Broadcasting Services (TELSIKS 2009).
Like the previous conference, TELSIKS 2007, this year’s
event was organized jointly by the Faculty of Electronic Engineering Nis and the National Society for Microwave Technique and Technologies. TELSIKS 2009 was organized under
the technical co-sponsorship of the IEEE MTT Society, IEEE
AP Society, and IEEE Region 8, and in cooperation with the
IEEE Section of Serbia and Montenegro, IEEE MTT-S
Chapter, and IEEE Communications Society Chapter of Serbia and Montenegro, as well as the National Society for
Telecommunications and Society for ETRAN. The conference
was also supported by the Serbian Academy of Science and
Art, Academy of Engineering Science of Serbia and Montenegro, Ministry of Science and Technological Development, and
Ministry of Telecommunications and Information Society.
The main part of the conference program included presentations of scientific papers from a range of topics in the field
of telecommunications. The review of submitted papers was
done by an international board of conference reviewers, and
the final selection was made by the Conference Program
Committee. The total number of papers selected for presentation was 124, 19 of which were invited. Authors of the
papers are scientists from the following countries: Bosnia and
Herzegovina, Bulgaria, Canada, Czech Republic, Croatia,
Germany, Greece, Italy, Libya, Macedonia, the Netherlands,
Romania, Slovenia, Serbia, Ukraine, Spain, the United Kingdom, and the United States. There were 11 regular sessions
and five poster sessions. All papers scheduled for presentation
were published prior to the conference. The two-volume proceedings, accompanied by a CD-ROM, were distributed to all
registered participants. It is important to note that the Conference Proceedings, as an official IEEE publication, will be distributed by the IEEE Conference Publications Program
Prof. A. Marincic opening the TELSIKS 2009 Conference. From
left: Chairmen Prof. K. Rao, Prof. B. Milovanovic, Prof. S.
Tomacic, Prof. G. Stoyanov, and Prof. O. Fratu.
(IEEE CPP). All of the conference papers will be indexed in
the IET INSPEC database.
Like previous TELSIKS conferences, this year’s conference
was not just a conference with presentations of scientific
papers and exchange of experience, but also an event including many activities and meetings covering different important
issues related to the field of telecommunications. The first of
the additional activities was the Workshop “Trends in Multimedia Communications.” The lecturers were Prof. Kamisetty
R. Rao, University of Texas at Arlington, who gave a plenary
talk at the beginning of the Conference as well, and Prof.
Zoran Bojkovic, University of Belgrade, Serbia. Then there
were two roundtables organized during the conference: “Strategy of Scientific and Technological Development in tbe
Republic of Serbia 2009–2014” and “Accreditation in Higher
Education — Results and Forthcoming Activities.” Since actual Serbian topics were considered, both roundtables were held
(Continued on Newsletter page 4)
The Spanish ICT Hyper-Sector Reduces Its Activity by Seven Percent
By Juan Pedro Muñoz-Gea and Josemaría Malgosa-Sanahuja
Polytechnic University of Cartagena, Spain
The Spanish Association of Electronics, Information Technologies and Telecommunications (AETIC) has presented the
data of the ICT hyper-sector (telecommunications industries,
telecommunications services, information technologies, consumer electronics, electronic components, and professional
electronics) relating to the period between July 2008 and June
2009, which shows a reduction of 7 percent in activity. This
figure shows an abrupt change of trend in the ICT sector,
which has always starred in annual increments upward, except
in 2008, when zero growth was recorded.
All sectors that make up the hyper-sector have recorded
losses for the first time, highlighting the addition to this negative trend of the telecommunications services and information
technologies sectors (–4 and –2 percent, respectively). In addition, the second quarter of 2009 showed a rapid decline of 13
percent in the hyper-sector (compared to the same period in
2008) with rates ranging from –8 percent in telecommunications services to –38 percent in electronic components.
The telecommunication industries sector registered the
biggest fall of the hyper-sector from July 2008 to June 2009
(–25 percent), which accelerated in the second quarter of 2009
with a decline of 31 percent. Among other factors, this sharp
fall is associated with a decline (for the first time in Spain) in
the turnover of operators, a reduction in demand for fixed
networks, and stagnation of the fixed broadband market. In
the second quarter computers and networks fell by 40 percent,
while mobile terminals fell 17 percent in the first six months
of 2009.
The consumer electronics sector underwent the second
biggest decline, with a reduction in turnover of 22 percent
(–27 percent, second quarter). The main causes are focused
on the falling prices of TVs — around 25 percent — and a
reduction of close to 10 percent of sold units in the market.
The electronic components sector reduced its activity by 17
percent due to negative performance in other sectors. During
the second quarter, the trend worsened to –38 percent. The
professional electronics sector fell 8 percent, while from April
to June 2009 the decline was more pronounced (–15%)
because companies had exhausted the sector’s backlog, and
replacement was very limited.
The telecommunications services sector decreased (for the
first time in history) 4 percent in billing. The second quarter
figures went down four more points(–8%). All components
(Continued on Newsletter page 4)
Global Communications Newsletter • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
3
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
TELSIKS CONFERENCE/continued from page 3
in Serbian. Both of them gained the attention of the Serbian
academic community.
Since TELSIKS is dedicated to the introduction of new
telecommunications technologies and services, experimental
broadcasting of digital television programs in Nis was set up
during the Conference, in cooperation with the national
telecommunication company Telekom Srbija, as well as three
television companies, one of which has national coverage.
In addition to the activities related to the presentation of
scientific and professional results and activities, there were
several commercial presentations. Some of the leading information and communications technology (ICT) companies presented their latest solutions and products. Furthermore, there
was a one-day exhibition of IET journals. As a special attraction, the presentation of the project “Computer Simulation
and 3-D Modeling of the Original Patents of Nikola Tesla”
was held. It was organized by the Faculty of Electronic Engineering of Nis and the Nikola Tesla Museum in Belgrade.
Among the other activities, there were meetings of the
Executive Committee of the IEEE Section of Serbia and Montenegro, and the IEEE Women in Engineering Chapter of
Serbia and Montenegro. Furthermore, a meeting of the Serbian Society of Cable Operators’ representatives was held as
well.
The conference program was rich in social events as well.
There was a reception for invited authors organized by the
Mayor of the City of Nis, followed by a nice musical performance for all conference participants. A sightseeing tour for
all interested participants was organized. The conference gala
A
BEMaGS
F
dinner was organized at the end of the second day. Serbian
national cuisine, an orchestra performing Serbian and international music, and, most important, cheerful guests made the
evening memorable.
After the successful series of previous TELSIKS conferences, TELSIKS 2009 again offered a forum for promoting,
discussing, and spreading ideas in the field of telecommunications.
I would like to thank to all of the authors, members of the
TPC and organizing committee, reviewers, sponsors,
exhibitors, and all the others who participated in organization
and in the conference itself.
SPANISH ICT HYPER-SECTOR/continued from page 3
showed negative trends in income (fixed, mobile, pay TV,
wholesale services), except for Internet access, which had a
positive but declining rate. This situation is a direct result of
the combined effect of the fall in traffic and average revenue
per line and per minute, in both fixed and mobile telephony.
Despite all of this, mobile line and Internet penetration grew
moderately, and fixed access penetration was steady.
The information technologies sector has decreased by 2
percent in turnovers during the past 12 months (–15 percent
second quarter). The negative behavior of the software and
computer services has added to the behavior registered by
hardware equipment for a long time. This is due primarily to
decreased investment in professional systems, both public and
private.
NANJING CHAPTER/continued from page 1
Global
Newsletter
www.comsoc.org/pubs/gcn
STEFANO BREGNI
Editor
Politecnico di Milano - Dept. of Electronics and Information
Piazza Leonardo da Vinci 32, 20133 MILANO MI, Italy
Ph.: +39-02-2399.3503 - Fax: +39-02-2399.3413
Email: bregni@elet.polimi.it,
s.bregni@ieee.org
___________ __________
scholarship students from all over the world, about half of
them IEEE members, participated in the conference. WCSP
2009 was very successful, and conference attendees felt that
the conference provided them with a platform where they
could learn about new advances in the wireless communications and signal processing fields, and meet and discuss issues
in research with peers working in the same area. They also
expressed that they look forward to participating in WCSP
next year.
IEEE COMMUNICATIONS SOCIETY
KHALED B. LETAIEF, VICE-PRESIDENT CONFERENCES
SERGIO BENEDETTO, VICE-PRESIDENT MEMBER RELATIONS
JOSÉ-DAVID CELY, DIRECTOR OF LA REGION
GABE JAKOBSON, DIRECTOR OF NA REGION
TARIQ DURRANI, DIRECTOR OF EAME REGION
ZHISHENG NIU, DIRECTOR OF AP REGION
ROBERTO SARACCO, DIRECTOR OF SISTER AND RELATED SOCIETIES
REGIONAL CORRESPONDENTS WHO CONTRIBUTED TO THIS ISSUE
MILAN JANKOVIC, SERBIA (___________
LJILJAMJ@EUNET.YU)
JOSÉ MARIA MALGOSA-SANAHUJA, SPAIN (______________
JOSEM.MALGOSA@UPCT.ES)
EWELL TAN, SINGAPORE (____________
EWELL.TAN@IEEE.ORG)
®
A publication of the
IEEE Communications Society
4
Communications
IEEE
Global Communications Newsletter • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
October 4-6, National Institute of Standards and Technology
A
BEMaGS
F
www.ieee-smartgridcomm.org
1st IEEE International Conference on Smart Grid Communications
October 4 – 6, 2010 ~ National Institute of Standards and Technology (NIST)
Gaithersburg, Maryland, USA
Developing the Smart Grid has become an urgent global priority as its economic, environmental, and
societal benefit will be enjoyed by generations to come. Information and communications technologies are
at the core of the Smart Grid vision as they will empower today’s power grid with the capability of
supporting two-way energy and information flow, isolating and restoring power outages more quickly,
facilitating the integration of renewable energy sources into the grid and empowering the consumer with
tools for optimizing their energy consumption.
The 1st IEEE International Conference on Smart Grid Communications (SmartGridComm) is centered on all
communications aspects that are relevant to the Smart Grid and aims at bringing together researchers
from Academia, Industry, and National Labs to exchange novel ideas, explore enabling technologies,
discuss innovative designs, and share field trial experiences and lessons learnt.
The IEEE SmartGridComm Conference will be constituted of twelve Symposia. Each symposium will
address a particular aspect of Smart Grid Communications. Prospective authors are invited to submit
original contributions (standard two-column IEEE format and up to 6 pages) on all aspects of Smart Grid
Communications to one of the following Symposia:
Paper Submission Deadline: May 1, 2010
Camera Ready Paper Due: Aug. 1, 2010
Notification of Acceptance:
Author Registration Deadline:
July 19, 2010
Aug. 1, 2010
George Arnold, NIST, USA
Stefano Galli, Panasonic R&D, USA
Fred Baker, Cisco, USA
Hamid Gharavi, NIST, USA
Simon Haykin, McMaster University, Canada
www.ieee-smartgridcomm.org
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SERIES EDITORIAL
DESIGN AND IMPLEMENTATIONS SERIES V:
IMS APPLICATIONS AND SUPPORT
Sean Moore
T
he IEEE Communications Magazine editorial team characterizes Series and Feature Topics as covering either
vertical or horizontal topics. A topic is vertical if it is focused
on a particular communication technology (e.g., next-generation optical switching). A topic is horizontal if it focuses on
some operational aspect that is common across many communications technologies, such as interoperability testing or performance testing. The Design and Implementation (D&I)
Series is horizontal, as D&I articles are intended to teach
industry professionals about building next-generation communications products and services, regardless of the specific technology area.
Coincidentally, however, five of the six articles in this fifth
installment of the D&I Series discuss IP multimedia subsystem (IMS) technology or support for IMS business operations.
These articles were submitted independently without a call or
solicitations for IMS-related topics. This coincidence indicates
the market strength of IMS and suggests that IMS is in an
active stage of its R&D life cycle where the graph of R&D
investment vs. time has a positive second derivative. Hence,
this installment of D&I has been verticalized around IMS and
may be used by Communications Society members as a
resource to improve the efficiency and quality of their IMSrelated projects.
IMS is an architectural framework that uses Session Initiation Protocol (SIP) signaling to deliver multimedia services over IP infrastructure. IMS was developed by the
Third Generation Partnership Project (3GPP) standards
organization specifically to support delivery of 3G mobile
services. The IMS framework is sufficiently flexible such
that it is now being used as the basis for both service
provider and enterprise solutions using both wireless and
wired infrastructure.
At the core of an IMS implementation is intelligent SIP
routing, which interconnects users, devices, services, and
applications. The capability to rout a SIP message to multiple
applications, which are hosted by Java Enterprise Editioncompliant application servers (JEE A/S), may be the most
strategic capability of IMS. For example, when a caller sends
a SIP INVITE message to a callee to create a session, the
IMS SIP router routs the INVITE message through a chain of
independent applications or services for, say, call screening,
billing, QoS assurance, presence, user location tracking, E911,
contact center information, contextual services, sleep proxies,
or enterprise applications. These applications are hosted by a
JEE A/S. Thus, unlike the call signal processing interfaces of
24
Communications
IEEE
legacy softswitches, which in practice were accessible only by
telephony experts of the operating companies, an IMS-based
implementation is truly an open architecture that is accessible
by the large community of Java enterprise application and
Java web application developers. The IMS framework converges telephony applications and enterprise IT applications
by providing a gateway (in the form of the SIP IMS service
control [ISC] interface) between the signaling protocols of
telephony systems and those of IT systems.
The possibilities for high-value converged IMS applications
and services are seemingly endless. But IMS service providers
and enterprise vendors may not have the R&D resources necessary to meet their customers’ demands for IMS technologies
— particularly the need for Java enterprise applications —
and are therefore dependent on third parties to provide the
resources. These third parties in some cases may be the customers themselves. IMS’s open architecture enables thirdparty participation, but IMS service providers and vendors
must assist in this process and compete for these third-party
resources by providing effective application development tools
and educational materials to the third party markets. Currently, these market forces are quite strong, in direct correlation
with the increasing penetration of IMS, and have clearly factored into the “coincidence” that so many of the D&I articles
in this issue are IMS-related.
Sal Loreto and an Ericsson team contributed “IMS Service
Development API and Testbed,” which addresses a significant
obstacle to IMS application development for Java developers.
In a native IMS implementation, SIP is used for communications between the SIP router and JEE A/S-hosted applications, which requires that the application be structured as a
SIP servlet to handle and process SIP messages. The Java
application developer must then work directly with the SIP
protocol instead of with a programmatic Java library, which is
contrary to the expectations of the Java development community. This article discusses how to design and build a Java programmatic interface that abstracts IMS application
development away from the SIP protocol level. A similar
approach is used by other IMS vendors and for the same reason: to make IMS application development highly accessible
to enterprise Java developers.
Alejandro Cadenas and Alejandro Sanchez-Esguevillas of
Telefonica I+D, working with Belen Carro of the University
of Valladolid, describe a practical deployment of contextual
services used to intelligently route calls in an IMS implementation (“Deployment of Contextual Corporate Telco Ser-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SERIES EDITORIAL
vices”). Context-aware routing factors the users’ context —
location, mobility mode, presence state, current role, and so
on — into call routing decisions. In the enterprise setting,
context-aware routing may be used, for example, to route a
call to the current best contact device while selecting the most
appropriate media (voice, voice mail, text, email, etc.) for the
communication. In the consumer setting, context-aware routing may be used to satisfy immediate consumer needs; for
instance, a subscriber desiring to exchange currency may be
routed to the closest bank providing an exchange service. The
authors show how to architect context services in the IMS
framework to maximize the efficiency and effectiveness of
context-aware routing.
The demand for contextual services drives a correlated
demand for sensor networks to collect context data. In a
serendipitous complement to the article by Cardenas et al.,
May El Barachi and a team from Concordia University and
United Arab Emirates University describe how to integrate
IMS implementations with wireless sensor networks (WSNs)
for collecting contextual data (“Architectural Components for
the Integration of IMS and WSN”). The design and implementation includes a WSN/IMS gateway and an extended
presence service for managing contextual information. Barachi
et al. prove out their solutions by creating IMS applications
that use the contextual information collected from the WSNs.
Any commercial success of IMS may be attributed in part
to technology benefits, but the majority share of the credit
should be given to the business operations. Without efficient
marketing, planning, delivery, and operational support for
IMS services, IMS service businesses may not succeed. The
next two articles provide valuable information that will help
improve the chances for success.
Often, communications services initiatives, such as an IMS
services rollout, are launched with (often unfounded) goals to
achieve some percentage of market penetration, such as “near
100 percent!” Is there a simple, low-cost method for estimating realistic goals? Or equivalently, is there a method to estimate if the current business environment supports the
attainment of such goals, or are additional stimuli needed?
Ryszard Struzak, IEEE Life Fellow, proposes such a method
in “Broadband Internet in EU Countries: Limits to Growth.”
Struzak borrows the logistic growth function used in biological
system modeling and applies it to early data from the European i2010 broadband initiative, which has a goal of 90 percent broadband penetration for each member country of the
European Union. Given current conditions, Struzak’s method
shows that many countries are not on a trajectory to achieve
the goal, which suggests that actions be taken to change the
business environment and thereby improve the likelihood that
the i2010 goals will be met. Struzak’s method is quite general,
and its application to analysis of IMS service penetration
goals should be straightforward.
IMS not only drives more and richer IP multimedia services but also drives growth of organizations that provide IP
multimedia services over third-party IP network infrastructure.
The network operators therefore need to measure precisely
the traffic volumes of individual media services for accurate
cost accounting and charging. Operators must also measure
traffic to ensure QoS agreements, and to use the data in network design and planning. JungYul Choi and a team from
Korea Telecom built a service traffic management system to
solve these problems. In “Service Traffic Management System
for Multiservice IP Networks,” Choi et al. share their experience and lessons learned on the project, prove empirically the
accuracy of their methods, and discuss how their system supports the new business and management processes of Korea
Telecom.
Prasant Misra and a team from the University of New
South Wales and the CSIRO ICT Centre share their experience with designing and building communication systems that
may be deployed, for example, in disaster-response situations,
urban war zones, and other high-stress environments (“Safety
Assurance and Rescue Communications in High-Stress Environments”). Real-world testing of communications systems in
such environments is obviously a barrier, but underground
mines display many of the extreme communications characteristics of high-stress environments and may proxy for them as a
design and implementation testbed. Solving the underground
mine communications problem is by itself a significant contribution, but such solutions may be leveraged into other highstress environments. Misra et al. provide a tutorial on extreme
communications characteristics and how they may be
addressed, discuss current solutions for communications in
mines, and conclude with an experiment assessing the performance of a wireless sensor network in an underground mine.
I hope you both enjoy and learn from this fifth installment
of D&I as much as I did in serving as its editor. If these articles and past installments of D&I have inspired you to consider contributing your valuable D&I knowledge to the
Communications Society’s members, contact me directly
(smoore-phd@ieee.org)
_______________ so that together we can create a firstrate publication for the benefit of our industry members. The
D&I Series Call for Papers may be found at ___________
http://www.comsoc.org/livepubs/ci1/info/cfp/cfpcommagdesignimplementa________________________________________
tion1.htm.
______
Please join me in acknowledging everyone that made possible this installment of the D&I Series: the authors’ sponsoring
organizations for donating their time, Avaya for donating my
time, the Editor-in-Chief Steve Gorshe, Joe Milizzo and Jennifer Porcello and her team at IEEE Communications Magazine, and the many anonymous reviewers who ensured the
quality of the articles.
BIOGRAPHY
_____________ has over 25 years of
S EAN MOORE [M’01, SM’03] (smoore-phd@ieee.org)
experience in a variety of technology industries, and has been working in
networking and telecommunications since 2001. He currently works at
Avaya, a vendor of enterprise communications solutions, where he has
served as chief scientist in the gateways division, as an enterprise architect
in the CTO office, and currently serves as the chief architect for developer
platforms and as Avaya's representative to ECMA International. One of his
current projects is architecting the software development kits (SDKs) and
platforms for use by third-party developers to create enterprise telephony
web applications and SIP/IMS applications. In the past he was chief scientist at Cetacean Networks, a vendor of advanced routers and routing applications, senior director of R&D at MadeToOrder.com, a developer of
supply-chain management e-commerce solutions, and director of advanced
systems and director of business development at BBN Corporation, an R&D
services provider to the U.S. Department of Defense. He has made contributions in peer-to-peer media distribution protocols, network admission
control, web-telephony convergence, network tomography, digital signal
processing, fast Fourier transform (FFT) algorithms, medical imaging, global
climate modeling, global-scale distributed databases, global-scale logistics
and scheduling systems, e-commerce, genetic algorithms, automated hardware design, queuing theory, TCP technology, and scheduling theory. His
software for FFTs on the 2-sphere is open-sourced as SpharmonicKit. He
also serves as the Design and Implementation Series Editor and Associate
Editor-in-Chief for IEEE Communications Magazine. He is a charter member
of the IEEE Standards Activities Board. He received a B.S. degree in electrical engineering from Tulane University in 1983, an M.S. in mathematics
from the University of New Orleans in 1990 (SIAM Applied Mathematics
Award co-recipient), and M.S. and Ph.D. degrees in computer science from
Dartmouth College in 1993 and 1994, respectively.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
25
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
IMS Service Development
API and Testbed
Salvatore Loreto, Tomas Mecklin, Miljenko Opsenica, and Heidi-Maria Rissanen, Ericsson Research
ABSTRACT
The IP multimedia subsystem defined by the
Third Generation Partnership Project is the
architecture merging the Internet and telecom
worlds. The IMS was designed to make it easy
for third-party developers to make their applications available to all IMS users, and by doing so
provide more than only the basic telecom services like voice, messaging, presence and contact
management. However, good knowledge of the
IMS network architecture and the underlying
Internet protocols is still needed to develop IMS
applications. In addition, telecom expertise is
needed to deploy the application and provision
users. To ease the development and deployment
process, it is essential to provide application
developers with APIs and similar tools available
for Web 2.0 application development today. In
this article we explore the architectural and protocol aspects that enable third-party IMS application development and deployment. We also
study how the applications will coexist with other
applications already deployed in the IMS. Moreover, we describe Java libraries exploiting the
functionality of the IMS both in the terminal
client and within the core network. We also
show how these Java libraries can be used for
developing and deploying new applications in an
IMS testbed, which provides IMS functionality
over commercial 3G networks.
INTRODUCTION
The Third Generation Partnership Project
(3GPP) developed the IP multimedia subsystem
(IMS) [1] architecture with the aim of providing
and handling a large variety of innovative services. To achieve these goals the IMS uses Internet Engineering Task Force (IETF) protocols.
In particular, it uses Session Initiation Protocol
(SIP) [2] as the session control protocol to establish and control multimedia sessions.
IMS is a SIP-based network architecture that
provides a multiservice environment with multimedia capabilities. For this reason, 3GPP has
introduced logical elements as well as protocol
mechanisms that are not defined in the plain
IETF SIP.
The IMS network contains at least one home
subscriber server (HSS), which is the central
repository for user-related information (e.g., the
26
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
service to which the user is subscribed). As logical elements IMS architecture defines different
call session control functions (CSCFs), which are
essentially SIP proxy servers supporting IMS
specific mechanisms. The serving CSCF (SCSCF) is the central node of the signaling plane.
It is essentially a SIP server, but also performs
session control. It also acts as a SIP registrar,
maintaining a binding between the user’s location and the user’s SIP address of record. All the
SIP signaling that is sent or received by the IMS
terminal traverses the allocated S-CSCF. The SCSCF inspects every SIP message and determines whether the SIP signaling should visit one
or more application servers (ASs) en route
toward the final destination using the information contained in the initial filter criteria (IFC).
This is illustrated in Fig. 1.
An AS is a logical IMS element that hosts,
executes, and provides the business logic for
end-user services. The AS can be located either
in the home network or in an external thirdparty network with which the home operator
maintains a service agreement. Thus, unlike IP
networks, an IMS network does not respect the
end-to-end Internet philosophy which mandates
that all the intelligence is at the edges of the
network. Instead, an IMS network allows part of
the intelligence of a service to be hosted inside
the network (i.e., in the AS). This way, terminals
with limited capabilities can access complex services using the support from the network.
The IMS architecture was designed to be flexible and expandable, allowing third-party developers to bring new services online. However, 3GPP
gave little attention to the issue of actually creating and deploying services easily. As a consequence, exposed IMS interfaces have been
complex and have not attracted developers. Therefore, the existing IMS services have been used just
like a set of operator controlled bit-pipe utilities.
During the last few years there have been
several efforts to expose IMS functionality to
application developers via high-level interfaces.
However, in order to provide new services to
users of the IMS network, it is also necessary to
implement the business logic for the new services within the network. Until now there has
not been any effort to expose IMS network capabilities to service developers via high-level interfaces that would have allowed easy development
and deployment of new services to an AS.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
To try to resolve this limitation we have published high-level Java IMS application programming interfaces (APIs) and an IMS testbed that
enables application developers, with no prior
knowledge about IMS or SIP, to build IMS
applications and deploy the applications in a live
IMS network (http://labs.ericsson.com).
The remainder of the article is organized as
follows. In the next section we explore the currently available Java APIs for both the terminal and
server sides, discuss their functionality, and highlight the inconsistencies and disalignments
between them. We then describe a presence service implementation built on top of the provided
IMS API. We discuss application and service routing in IMS. We show how the provided Java
libraries can be used for deploying new applications in an IMS testbed, which provides the IMS
functionality over commercial 3G networks. To be
able to use the services of the testbed network, the
user has to be provisioned to the network. This is
described later. In the final section we conclude
the article and propose some future work.
IMS CORE API
Java Micro Edition (Java ME) is the most ubiquitous application platform for cellular phones with
constrained resources. To lower the barriers for
Java ME IMS application development, the Java
Community Process (JCP) defined an IMS Services API package for the Java ME platform. This
package is provided in Java Specification Request
(JSR) 281 (http://jcp.org/en/jsr/detail?id=281). To
target the whole Java ME user community, JSR
281 was designed using the following design
principles:
• High abstraction level of the API allows
non-IMS developers to create IMS applications.
• Creating an IMS-aware application takes
only a few steps. Thus, the developer can
concentrate on the logic of the application.
• The API allows access to low-level APIs for
developers with IMS knowledge. On the
lower level, for example, JSR 180 (http://jcp.
org/en/jsr/detail?id=180), which is the SIP
API for Java ME, can be used.
The IMS API defines a set of service methods that can be used to build IMS applications:
• Session represents media exchange
between two IMS endpoints.
• Capabilities instance queries a remote
endpoint about its capabilities.
• Reference is used for referring to a
remote endpoint of a third-party user or
service.
• Subscription subscribes to event state
from a remote endpoint.
• Publication publishes event state to a
remote endpoint.
• PageMessage is used for instant messages
or exchange of small amounts of content
outside of a session.
No similar high-level IMS API has been
defined on the server side for Java Enterprise
Edition (Java EE). For Java EE there is JSR
289, which is the SIP Servlet API 1.1 (http://jcp.
org/en/
jsr/detail?id=289). It standardizes the
_________________
platform for delivering functionality for SIP sig-
AS
AS
IEEE
BEMaGS
F
AS
ISC
HSS
P-CSCF
S-CSCF
I-CSCF
Figure 1. SIP routing in IMS.
naling for Java EE applications.
SIP servlets are server-side components that
perform SIP signaling. They are managed by a
SIP servlet container, which is usually part of a
SIP AS. The SIP Servlet API takes care of managing network listening points, retransmissions,
and SIP message headers such as CSeq and CallID. The SIP Servlet 1.1 API also provides a
standardized way to develop converged applications that use both SIP servlet components and
standard Java EE components such as Enterprise Java Beans (EJBs).
For non-SIP experts to develop IMS applications the abstraction level of the JSR 289 API is
too low. Thus, a higher-level interface similar to
the terminal side’s JSR 281 is needed. To solve
this inconsistency between client and server
APIs, we implemented an IMS Core API for
Java EE.
IMS API LAYERS
The IMS core API for Java EE was designed to
also provide on the server side a higher level of
abstraction where the developer does not need
to be aware of the SIP Servlet programming
model or SIP signaling. As the SIP protocol is
asynchronous, communication initiated by
remote endpoints is handled by attaching listeners to the created services. Applications initially
need to create an ImsCoreService as a handle
for creating new SIP transactions and receiving
incoming transactions. A listener is attached to
the ImsCoreService to listen to remotely initiated transactions such as session invitations. When
the application needs to initiate a transaction,
the specific service is created by using the
ImsCoreService. As with attaching a listener to
the ImsCoreService, a listener can also be
attached to the specific service to receive messages related to the service. Notifications about
delivery success are delivered using the listener
attached to the specific service.
One of the design requirements for the IMS
core API was to have a similar IMS API for
both Java ME and Java EE. To achieve this, we
did alignments in the abstraction level, application, and service routing, and supported features. Full alignment is not possible due to the
differences between the Java platforms. The
IMS API layers for both Java EE (server side)
and Java ME (client side) are shown in Fig. 2.
On top of the IMS core API, it is possible to
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
27
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Server API layers
Registration to the
IMS network can be
done both from the
server and the client
BEMaGS
F
Terminal API layers
Application
Application
Application
Application
Application layer
A
Application
Application
Application
Application
side. However,
an application
IMS CoSe
Presence
Gropu mgmt
Presence
domain without the
Messaging
be in a trusted
IMS CoSe
Conformance
deployed on the
AS is considered to
need for registering
the user of the
application.
Media
control
JSR309
IMS Core API
JSR289
Java EE
(Sailfin application server)
IMS adaptation layer
(JSR 281)
JSR180
Java ME (client platform)
Figure 2. The IMS API layers.
build communication services (CoSes): a set of
rules, procedures, and allowed media for a specific type of service. The most common CoSes
are presence, messaging, and conferencing.
Developers can build their applications using
either the IMS core API or CoSe APIs, or both.
Developers who are familiar with IMS and SIP
concepts can also access lower-level APIs.
For brevity, this article does not provide
details of the CoSe services or APIs.
Server Implementation — As shown in Fig. 2,
on the server side there is a Sailfin AS at the
bottom as the deployment platform. Sailfin contains a JSR 289 library. The server side IMS API
implementation is built on top of the JSR 289
API. The abstraction level and design philosophy of the server implementation are based on
the JSR 281 API.
In the server API we used service and callback methods similar to the ones used in the
JSR 281 API. The callback concept is based on
the listener and observer design patterns
(http://java.sun.com/developer/JDCTechTips/200
6/tt0113.html).
_________ Using the callback mechanism, a
user can set a listener for an invoked method to
receive corresponding events.
The core API’s connections to the IMS network are handled by JSR 289 and the Java platform.
Terminal Implementation — On the terminal
side, there is an implementation of the JSR 281
API as shown in Fig. 2. In addition to that, we
implemented an IMS framework, which controls
the connection to the IMS network. The IMS
framework also takes care of routing the incoming requests to the correct application.
The IMS framework can only be used in terminals that have support for JSR 180, which is
28
Communications
IEEE
the SIP API for Java ME (http://jcp.org/en/jsr/
A JSR 180 stack, which was
implemented as part of the project, can be used
if there is no native JSR 180 in the terminal. So
far we have used only Sony Ericsson phones to
test the APIs.
detail?id=180).
__________
IMS CORE API FUNCTIONALITY
The core API provides high-level interfaces that
can be accessed from both the CoSe and application layers. The level of abstraction is similar to
the abstraction level used in JSR 281. New IMS
services can be introduced by combining existing
interfaces or adding new ones. Currently exposed
functionality is described in the following.
Registration — To be able to use IMS functionality, the user needs to be registered to the
IMS network. Registration is done by invoking
the Registration interface method and by
listening to the registration events. Registration
to the IMS network can be done from both the
server and client sides. However, an application
deployed on the AS is considered to be in a
trusted domain without the need to register the
user of the application.
SIP Session Management — The SIP session
management interface provides the SIP core
protocol functionality. By invoking interface
methods and listening to the events, it is possible
to perform all basic SIP session actions: invite a
party to a session, accept and terminate a session, and respond to requests. SIP signaling is
taken care of by the API implementation, but
the user has to know how to manage concepts
such as an ImsSession and ImsMedia.
Instant Messaging — The IMS core API supports two modes of instant messaging: session
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
The presence service
Terminal
Application server
Java ME MIDlet
Java EE application
online and, if they
CoSe layer
Presence
informed whether or
not other users are
Application
interfaces
CoSe layer
allows a user to be
are online, whether
Presence
they are idle or busy,
as well as other
IMS adaptation
layer (JSR231)
The core API layer
(JSR281 alike)
JSR180
stack
Service
router
details of their
communication
means and
capabilities.
IMS
interfaces
JSR180
stack
Client Java ME
platform
Device
infrastructure
Application
router
ServerSailfin
platform (JSR289)
Network
infrastructure
Figure 3. Presence example of a CoSe service.
mode and page.
Page messaging uses SIP MESSAGE, as
defined in [3]. Due to the size limitations of IMS
nodes like the session border controller (SBC), a
bigger amount of content should preferably be
sent using the session mode.
Session mode messaging supports exchanging
arbitrarily sized content between users. In session mode the messaging service creates a session with the destination and delivers the
messages using the Message Session Relay Protocol (MSRP) [4]. An MSRP session is set up by
exchanging certain information using SIP, such
as the MSRP uniform resource identifier (URI).
SIP Event Notification and Publication — The
event notification and publication interface provides methods for using SIP subscriptions, notifications, and publications. SUBSCRIBE and
NOTIFY functionality is implemented according to
RFC 3265 [5] and PUBLISH according to RFC
3903 [6]. On top of this functionality, a CoSe service can, for example, publish a user’s presence
information or subscribe to other users’ presence
information.
PRESENCE SERVICE
IMPLEMENTATION
The presence service allows a user to be
informed whether or not other users are online
and, if they are online, whether they are idle or
busy, as well as other details of their communication means and capabilities. Capabilities can
be used to indicate if a user supports audio,
video, or instant messaging.
Presence service is the first CoSe service we
provide to developers, which is built on top of
the IMS core API in Java EE and on top of JSR
281 in Java ME. Examples of both a Java ME
MIDlet and a Java EE application using the
presence service are shown in Fig. 3.
At the time of implementation, JCP had not
defined any presence API; thus, our implementation is not based on any JSR but follows Open
Mobile Alliance (OMA) application-level specifications. The Presence API allows users to publish presence information and subscribe to other
users’ presence information. The API uses
Extensible Markup Language (XML) Configuration Access Protocol (XCAP) [7] for manipulating presence lists and presence authorization
rules on the XML Document Management Server (XDMS). Presence authorization rules define
the authorizing policy for watchers. Mechanisms
to create, modify, fetch, and delete XML documents are defined by OMA. To get real-time
updates about the changes in the XML documents, users can subscribe to the different events
by using SIP NOTIFY requests as defined in the
SIP event notification framework [5]. SIP event
notification and publish functionality is used
from the IMS core API in Java EE and from the
JSR 281 API in Java ME.
APPLICATION AND SERVICE ROUTING
As described earlier, IMS is an architecture able
to support several end-user services. The partic-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
29
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Once the chain of
applications is set for
the initial request,
all subsequent
requests will be
routed to the
applications in the
same order.
This way, complex
services can be built
without making
individual
applications
too complex.
ular service a user intends to use in a session has
to be identified at the time the IMS session is
established so the S-CSCF can trigger the SIP
signaling to the particular AS providing the service. This is illustrated in Fig. 1. SIP messages
have to carry a tag to allow the terminal and network to identify the service intended to be used
during an IMS session.
The SIP Caller Preference mechanism (RFC
3841) [8] describes a set of extensions to SIP
that allow a caller to express its preferences
about request handling in servers. These preferences include the ability to select to which URI
a request gets routed, and make it possible to
specify certain request handling directives in
proxies and redirect servers. Three request header fields, which specify the caller’s preferences,
are specified: Accept-Contact, Reject-Contact,
and Request-Disposition. In particular, the
Accept-Contact header explicitly contains the
desired properties of a terminal to which the
request is to be routed. 3GPP extends this
mechanism to also convey the communication
service identifier (ICSI) and the IMS application
reference identifier (IARI) in the Accept-Contact header.
As the logic of each standardized IMS service
can be utilized by a number of different service
applications, each of which implements a particular end-user service, both the ICSI and IARI
tags are needed in the SIP Accept-Contact header in our API implementation. These tags are
needed to determine the correct IMS services
and to address the correct application.
SERVICE ROUTING
As the IMS Core API supports the IMS multiservice concept where several services are available to the user, a service routing mechanism is
needed in the IMS Core API. To route the
request to the correct service, we use the 3GPP
ICSI feature tag in the SIP Accept-Contact
header. For example, multimedia telephony service (MMTel) would be identified with the following ICSI tag: *;+g.3gpp.app ref =
“urn:Aurn-xxx:3gppservice.ims.icsi.
mmtel”.
The ICSI tag is configured in the application deployment descriptor and registered in
the deployment phase, described later. Two
dependent tag components, ServiceID and ServiceName, are linked with a configured service
and used in subsequent service routing. Providing a combination of ServiceID and ServiceName allows application developers to
instantiate the same IMS service with different
configurations.
A service router (SR) is a core API component that handles routing logic on the server
side. The IMS system is responsible for routing
SIP signaling to the applications that are part of
a service. Routing is based on the first request
sent to the S-CSCF. The request initiator inserts
an ICSI tag into the Accept-Contact header and
sends it toward the IMS network. The S-CSCF
uses the IFC to identify the AS that has to be
included in the signaling path. After the request
is received at the AS container, the target application is selected using application routing,
which is described later. Service selection within
30
Communications
IEEE
A
BEMaGS
F
the application is done by the SR. The SR uses
the received tag to select the service within the
application. If the requested service is identified,
the request is sent toward the service for further
processing.
APPLICATION ROUTING
A complete service, such as a pure SIP or converged service, can be built by combining several
applications hosted by an AS. Each individual
application performs a part of the service independent of the others. The SIP container needs
to determine which application to invoke when
an initial request is received. This is done by an
application router (AR), which implements the
SIPApplicationRouter interface defined in JSR
289. The AR is not a part of the container, but
rather an extension thereof.
The AR may use any algorithm or data source
to determine the order in which applications are
invoked for an initial request. Once the chain of
applications is set for the initial request, all subsequent requests will be routed to the applications in the same order. This way, complex
services can be built without making individual
applications too complex. The application independence and composition used in JSR 289 is
adapted from the distributed feature composition (DFC) architecture [9].
Converged applications are identified by the
application name, defined either in the deployment descriptor or as an annotation, as described
later. When an application is deployed or undeployed, the AR is notified about the change in
the available applications. At the start, the container notifies the AR about the available applications. The names of the affected applications
are included in the notifications.
APPLICATION ROUTER IMPLEMENTATION
In our AR implementation we use the feature
tag defined in RFC 3840 [10] and RFC 3841 [8]
to map initial requests to the correct application.
We did not implement the required features for
application composition in the AR, since only
basic services were needed.
When the container receives an initial request
of which it has no prior knowledge, it forwards
the request to the AR. The AR is not allowed to
modify the request in any way, but it extracts the
required information from the request to be able
to select the correct application. In our implementation we extract the Accept-Contact header
and analyze its content. The AR will extract the
IARI value to select the application to which the
request should be routed. The ICSI is not used
by our AR implementation, but can be used by
the SR, as described earlier.
DEPLOYMENT OF APPLICATIONS
To deploy an application using the IMS APIs to
a real IMS network, the IMS testbed provided
by Ericsson Labs can be used (http://developer.
labs.ericsson.net/apis/mjcf). In this testbed the
SIP servlet container is a Sailfin AS (https://sail_______
fin.dev.java.net). Sailfin is based on the Glass___________
FishAS (https://glassfish.dev.java.net) with a SIP
servlets technology extension. The network
architecture of the testbed is shown in Fig. 4.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
The network contains an S-CSCF, HSS, SIPAS,
and a presence server containing an XDMS. The
Sailfin AS is located in another network that is
connected to the IMS network with a virtual private network (VPN) tunnel. An SBC manages
the signaling and MSRP streams to and from the
terminals. There is also a provisioning server,
which is discussed in the next section.
To deploy an application to the IMS network,
the application needs to be uploaded to the Sailfin AS. Deployment of IMS applications is similar to deployment of normal web applications.
Directories and files of an IMS application
should be laid out according to the standard web
application layout.
A standard web application configuration file,
sun-web.xml, is needed to deploy the SIP application. If HTTP servlets are used, a web.xml
deployment descriptor is also needed. In addition, the following SIP-specific configuration
files are needed:
• sip.xml is the SIP deployment descriptor.
It contains the name of the application and
servlet mappings. It is similar to the standard web.xml file, but instead of a URL
pattern used in web.xml, the header field of
a SIP request is used.
• ims.xml is the IMS descriptor of the application. It is not part of the SIP servlet API,
but a descriptor needed by the IMS APIs.
ims.xml contains the ID of the service, the
address of the CSCF and the SIP URI of
the user on whose behalf the AS is performing actions.
As an example, we have an application named
com.imsinnovation.sendmessage, which
is a simple application able to receive and send
page messages. The name of the application is
defined in the sip.xml deployment descriptor:
<?xml version=”1.0” encoding=”UTF-8”?>
<sip-app>
<app-name>com.imsinnovation.sendmessage</app-name>
<display-name>Page message test</display-name>
The corresponding IARI value to route initial
request to this application is
+g.3gpp.app_ref=
”urn:urn-xxx:3gpp-application.
com.imsinnovation.sendmessage”
PROVISIONING
Before the user can use the services of the IMS
network, s/he needs to be provisioned to the network. In our implementation we use the Ericsson Multi Activation (EMA) provisioning system
(http://www.ericsson.com/solutions/page.asp?Arti
c________________________________
leId=89A4A5D2-13F7-477C-9636-CB92
________ to provision users to the IMS netFA37FD6F)
work. EMA provides a provisioning interface
called the Customer Administration Interface
Third Generation (CAI3G). CAI3G follows
industry standards and best practices, and it uses
Simple Object Access Protocol (SOAP) over
HTTP. An EMA system can be used independent of the operator.
XCAP
SIP
Diameter
HTTP/CAI3G
LDAP
IEEE
BEMaGS
F
Sailfin AS
Internet
Terminal
SBC
NAT/FW
Provisioning
server
(EMA)
S-CSCF
Presence
server
HSS
Figure 4. Ericsson Labs IMS network.
To contact the EMA provisioning system
located in the Ericsson Laboratories IMS testbed
network, we have implemented a provisioning
application. This application is deployed on the
Sailfin AS in the testbed network. Data related
to the applications is stored on the AS, and data
related to the services of the IMS is stored on
the HSS. To make an initial provisioning request,
a SOAP-based authorization request is sent from
the provisioning application to the provisioning
server. This request contains the mobile subscriber integrated services digital network
(ISDN; MSISDN) identifier of the user.
If the response is successful, the user will
receive an SMS message with a provisioning
password to the mobile phone. To finalize provisioning, a final request containing the password
is sent from the application to the provisioning
server. This final request creates the IFC, and
provisions the user to the IMS network and its
services. CAI3G provides the possibility to later
update the user profile and for example to provision new services.
LESSONS LEARNED AND
FUTURE WORK
The IMS vision is to create an ecosystem where
all the applications provided by the developer
community are available for all users of the IMS.
As IMS networks are being installed in operators’ networks around the world, it seems essential to have APIs and tools available for IMS
application developers similar to the tools available for web developers today. We address this
problem by providing developers with high-level
Java APIs exposing the functionality of the IMS
network and a live IMS testbed network.
The main objective of this project was to publish high-level Java EE and ME APIs that would
allow web application developers to include IMS
functionality in their applications. One of the
requirements was that no prior knowledge about
IMS or SIP should be required of the developers. Thus far we have published a core API pro-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
31
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
To efficiently use
presence
information, some
information about
presence standards
could be useful.
Defining good
samples on how to
use these concepts
will help the
developers to build
their own
applications.
32
Communications
IEEE
viding the functionality described earlier. We
have also published a presence API, which is an
example of a CoSe service, for both Java EE and
Java ME.
As we are trying to merge the telecom and
Internet worlds, we are dealing on one side with
asynchronous network signaling and on the
other with synchronous Internet signaling. One
of the biggest difficulties came from differences
between SIP and HTTP sessions. Although both
protocols use a request-response pattern and
their syntaxes are similar, SIP is still an asynchronous communications protocol. As communication systems are inherently stateful, state
and transaction management is arguably the
biggest shortcoming of the SIP servlet layer we
have to consider. To balance the two different
approaches, we introduced some level of state
information in between and used a callback
mechanism dependent on network interactions.
For balancing on the Internet side we used
Comet’s (http://docs.sun.com/app/docs/doc/
8204496/ghgxk?a=view) asynchronous, non_________________
blocking HTTP mechanisms.
Another difficulty we faced during the project
was local testing of IMS applications. Although
we had an IMS environment for local testing, it
was not fully compliant with standards and
requirements used in the real testbed network.
This meant it was not possible to test IMS network-dependent functions locally but in the
remote, real IMS network. Therefore, testing of
functions like presence or provisioning was done
in two steps.
Most of the developers who have used the
APIs thus far have been students. The abstraction level in the APIs is higher than SIP, and
that has been clear enough for most of the
developers. However, to efficiently use presence information, some information about
presence standards could be useful. Defining
good samples on how to use these concepts
will help the developers build their own applications.
It is clear that providing only Java APIs is not
enough for the needs of the developers. Our
current and future work is related to implementing interfaces other than Java. Currently, we are
implementing APIs (e.g., for representational
state ttransfer [REST]). Other interfaces such as
.Net or SOAP need to be considered as well. In
addition, real-time media such as voice and
video are not yet supported, but implementation
work is ongoing and follows the JSR 309 Media
Server Control API (http://jcp.org/en/jsr/
detail?id=309). Support for streaming media is
__________
still an open issue since the current Java ME
platform does not support it.
A
BEMaGS
F
For the sake of brevity, this article does not
provide details of the CoSe services or APIs;
however, we intend to publish these details at a
future date.
REFERENCES
[1] 3GPP TS 23.228, “IP Multimedia Subsystem (IMS); Stage
2,” Sept. 2008.
[2] J. Rosenberg et al., “SIP: Session Initiation Protocol,”
IETF RFC 3261, June 2002.
[3] B. Campbell et al., “Session Initiation Protocol (SIP)
Extension for Instant Messaging,” IETF RFC 3428, Dec.
2002.
[4] B. Campbell, R. Mahy, and C. Jennings, “The Message
Session Relay Protocol (MSRP),” IETF RFC 4975, Sept.
2007.
[5] A. B. Roach, “Session Initiation Protocol (SIP) — Specific
Event Notification,” IETF RFC 3265, June 2002.
[6] A. Niemi, “Session Initiation Protocol (SIP) Extension for
Event State Publication,” IETF RFC 3903, Oct. 2004.
[7] J. Rosenberg, “The Extensible Markup Language (XML)
Configuration Access Protocol (XCAP),” IETF RFC 4825,
May 2007.
[8] J. Rosenberg, H. Schulzrinne, and P. Kyzivat, “Caller
Preferences for the Session Initiation Protocol (SIP),”
IETF RFC 3841), Aug. 2004.
[9] M. Jackson and P. Zave, “Distributed Feature Composition: A Virtual Architecture for Telecommunications Services,” IEEE Trans. Software Eng., 1998.
[10] J. Rosenberg, H. Schulzrinne, and P. Kyzivat, “Indicating User Agent Capabilities in the Session Initiation Protocol (SIP),” IETF RFC 3840, Aug. 2004.
BIOGRAPHIES
S ALVATORE L ORETO [SM] received his degree in computer
engineering from the University of Napoli Federico II, Italy,
in 1999. He obtained a Ph.D.degree in computer networks
in 2006. Since 2000 he has been working for Ericsson, first
in Italy and now in Finland. He is currently working as a
research scientist at Ericsson Research in Finland. He has
authored and co-authored several papers on transport and
signaling protocols. His research interests include signaling,
multimedia applications, transport protocols, and network
security. He is also an active member of the IETF and currently co-chairs the HyBi wg.
T OMAS M ECKLIN has been working for Ericsson since 1993
with various communication technologies. He graduated in
1994 from the Computer Science Department of Tekniska
Lroverket i Helsingfors. Since 1999 he has been working as
an architect for a number of SIP-based call controllers for
IMS. Currently he is working as research scientist at Ericsson
Research, Finland (Nomadiclab), with his main focus on service enablers, multimedia applications, privacy, and security.
M ILJENKO O PSENICA received his M.Sc. degree in electrical
engineering and information technology from the University of Zagreb — FER, Croatia, in 2001. He joined Ericsson in
1998. He has been working on a number of signalingbased call controllers. Currently he is working as a research
scientist at Ericsson Research, Finland (Nomadiclab), with
his main focus on service enablers, multimedia applications, privacy, and security.
HEIDI-MARIA RISSANEN received her M.Sc. degree in communications engineering from Helsinki University of Technology, Finland, in 2007. She joined Ericsson Finland in 2005.
Since 2006 she has been working as a research scientist at
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
______________
______________
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
Deployment of Contextual Corporate
Telco Services Based on Protocol
Adaptation in the NGN Environment
Alejandro Cadenas and Antonio Sánchez-Esguevillas, Telefónica I+D
Belén Carro, University of Valladolid
ABSTRACT
Deployment of contextual services is usually
constrained to specific areas where contextual
behavior can be obtained, mainly due to coverage
limitations of context acquisition devices. Although
end customers highly appreciate contextual services, those limitations make such services difficult
to commercialize. In this article we present a practical deployment of a contextual service offered by
a convergent telecommunications operator, whose
functionality is to provide intelligent context-based
call routing and rerouting, orchestrated from the
operator’s service layer. It is based on IMS control
layer capabilities to properly capture the situation
of the end user in a ubiquitous coverage area. The
user’s context is stored in a network-centric element in order to leverage that information across
different services, optimizing the system by
increasing the quality of the information captured
and processed. This implementation proves that
value-added contextual services may be built efficiently today using available products and protocols. Since contextual services will likely be a
valuable part of a service provider’s portfolio, this
implementation can help creators of new contextual services to meet cost and time-to-market
objectives.
INTRODUCTION
Context-aware applications are one of the new
paradigms in telecommunications commercial
environments, and significant research has been
done on this topic [1]. These applications are
based on appropriate sensing or detection mechanisms whose objective is to identify the situation, or context, of the user that will consume the
context-aware service. This sensing mechanism
can be either push or pull, depending on the specific sensor device and magnitude detected. Context is a diverse concept. It can range from the
physical presence of the user in a given room to
the emotional situation of a user that may affect
the way some services are delivered. It also has a
strong dependence on the time in which it is
acquired, as some types of information have
expiration times (location, activity, etc.).
34
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
The technology to capture such user information becomes instrumental for the deployment of
context-aware services. While detecting the presence of the user in a specific location may
require location capabilities from the access network, a global positioning system (GPS) on the
user terminal, or even a radio frequency identification (RFID) tag carried by the user, aspects
like the emotion or current activity of the user
are more difficult to capture or detect.
On top of that, even if the context of the user
can be properly detected, the sensor may be
deployed in a given location or area (e.g., an
enterprise building), and the context awareness
of the service can only be obtained for that area
in which the user information can be detected.
These issues become important barriers to
the successful deployment of commercial consumer context-aware services by telco operators,
since when targeting the mass market global coverage for the services is required.
The existing implementations of contextaware services are in several cases limited to laboratory environments or small areas [2].
Information about the user context is usually
processed by a service platform where the context-aware service is executed [3]. This platform
is connected to the user context sensing device.
Accordingly, in these scenarios a tight coupling
exists between the application and the sensors.
That is typically known as a vertical service layer
architecture in which all information required
for the service execution is obtained by and processed at the service execution platform, and no
sharing of user context information exists among
the different context-aware service platforms.
The telco operators have a relevant role on
such activities, especially the convergent operators (i.e., those operators that own both fixed
and mobile networks). This is due to the fact
that contextual services deployed by convergent
operators can efficiently make use of both fixed
and mobile access networks to process the user
information obtained by contextual sensors.
Such diversity of the access network means a significant increase in coverage of the contextaware service.
Hosting the context-aware platforms at the
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
telco operator service layer also offers advantages thanks to the sharing of user context information among different services. Therefore, the
captured contextual information can be reused
by any existing application in a horizontal way.
This means that the contextual information is
highly exploited across multiple applications.
Based on such objectives as horizontality of
the contextual information across the service
layer and the coverage range of the contextaware services, a telco contextual architecture is
presented in this article. This architecture uses
the capabilities provided by the IP multimedia
subsystem (IMS) convergent control layer specified by Third Generation Partnership Project
(3GPP) [4].
Details about the architecture design, protocols used, and a first service implemented and
deployed taking advantage of such capabilities
are also described. Different alternatives and
options chosen are provided, as well as details of
an implementation project of a context-aware
service. The article closes with lessons learnt and
conclusions.
IMS-BASED ARCHITECTURE:
CONTEXT ENABLER
Worldwide operators are currently evolving
toward network convergence as well as horizontal architectures of their service layers. Such evolution is driven by cost reduction and quality of
service (QoS) objectives, and is based on the
technology of IMS, whose main protocol is Session Initiation Protocol (SIP), specified by the
Internet Engineering Task Force (IETF) in
Request for Comments (RFC) 3261, which provides interworking capabilities with different
types of access networks.
Apart from other key advantages oriented to
security, authorization, accounting, and authentication (AAA), QoS, and so on, IMS is a suitable
framework to deploy the elements required to
handle the information about user context. In
IMS terminology, this is precisely the concept
called enabler [5].
An enabler is an element of the service layer
accessed through IMS. It is not a service itself,
but an entity whose objective is to provide additional information to the existing services or
applications at the service layer. An example is
the presence server [6], specified by the Open
Mobile Alliance (OMA). The presence server
stores and manages the presence information of
the end users, and provides presence information to any entity of the service layer that may
request it (with appropriate permissions).
Signaling exchanged between IMS enablers
and application servers located at the service
layer is based on SIP SUBSCRIBE/NOTIFY
methods, which provide a flexible and robust
procedure for the services to subscribe to the
updates of information handled by the enabler.
Therefore, a context enabler, handling information about the context of the user coming
from different sensors, fits perfectly in the IMS
standard architecture as opposed to the vertical
approach, in which each and every service will
handle the user context information in a sepa-
Application
servers
IEEE
BEMaGS
F
Enablers
Etc
Presence
Context
IMS
Access
network 1
Access
network 2
Figure 1. Target network architecture based on IMS, interoperable with different access networks.
rate way. In addition, the timing to introduce
such elements into the operator service domains
is appropriate, given that services compatible
with the IMS control layer are becoming commercial realities in the short or mid-term, and
such integration activities among different service platforms are currently taking place. The
IMS-based architecture with the context enabler
is presented in Fig. 1.
The main advantages of such anarchitecture
are the following:
Ubiquity of context-aware service: It can be
provided to the user regardless of geographical
location and type of device. This advantage is
based on the interoperability capabilities of the
IMS control layer. IMS interoperability [7] with
WiFi, cellular (3.5G, 3G, 2.5G), or fixed access
networks guarantees that any type of sensor used
to capture the user context or situation will be
able to notify the context enabler of such information for its storage and processing.
Enrichment of the contextual information:
The possibility of connecting a higher number of
sources of context information to the context
enabler will make the context information stored
at it more robust [8]. Different types of sensors
capturing different types of information (presence, activity, battery level, noise level, etc.) will
provide more accurate information about the
user.
Horizontal service layer: The fact that the
context information is stored at a single logical
point at the service layer enables complete sharing of the context information across all service
layer application servers. In order to achieve
that, services need to get subscribed to the context enabler for updates of the specific user context. The context information storage will be a
cluster of servers that will meet all requirements
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
35
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
of the project, and have driven many of the
implementation and design decisions.
Presence
DESIGN PHASE
SIP
IMS
PLMS
PSTN
OBG
PRI
IP
PRI
CTI
PBX
Figure 2. Designed architecture to implement, with the presence server and
OBG element. Both the mobile and fixed communications domains are
depicted.
of scalability and fault tolerance. Such information collection makes the development of contextual applications quicker and cheaper.
DESIGN AND IMPLEMENTATION
PHASES OF THE ARCHITECTURE
Context-aware services are suitable for different
market segments. However, the specific case
presented in this article is oriented to enterprise
users, with the following characteristics:
• End users (enterprise employees) usually
have different types of phones: corporate
mobile phone, office fixed desk phone, and
in some cases a corporate PC softphone
application.
• Enterprise users are more tolerant of learning curves and configuration of services
than the residential market.
• Specific development costs can eventually
be supported by the enterprise customer if
that means a significant enhancement of
the telecom service obtained by the enterprise.
Several parameters are considered critical
during design, implementation, and deployment
of services for enterprise segment. Those are
time to market, interoperability of the different
architectural elements, and development costs.
These key parameters have been the main focus
36
Communications
IEEE
During the design phase the enterprise telecom
system is analyzed. The global architecture is
depicted in Fig. 2, including both the mobile and
fixed domains, with the separate access networks, public land mobile network (PLMN) and
public switched telephony network (PSTN). The
corporate fixed communications are based on
customer premises equipment (CPE), which is
the private branch exchange (PBX) located at
the premises of each enterprise, but owned and
operated by the telco operator. There are a
number of corporate fixed lines (extensions)
connected to the PBX, typically one fixed line
per end user. The PBX will route the
incoming/outgoing calls through the external
connection, IP-based or via a primary rate interface (PRI) connection to the PSTN. In addition,
the PBX will also provide a number of enterprise communication services that are not usually available to residential and consumer
subscribers, such as call reject, call hold and
retrieve, call diversion, call transfer, and some
others. PC-based softphones may also get external access through the PBX. Eventually, a large
corporation may decide to manage their internal
communications. In that scenario the enterprise
still requires a PBX with external access provided by a telco operator, so the system implementation is exactly the same as if the internal
communications were also operated by an external telco operator. An additional element, the
operator business gateway (OBG), is also included in the architecture. This element is justified
in the implementation phase.
The corporate mobile communications are
based on a contract with the mobile telecom
operator, which provides mobile numbering
translation services when calls are made between
mobile phones of the same corporation (using a
short numbering scheme), in addition to others
very similar to the ones provided by the PBX on
the fixed side. Such services are usually provided
via intelligent network protocols from the service
layer at the mobile network.
In order to capture information on the situation of a user from a fixed phone, a trigger is
implemented at the PBX. That trigger converts
the PBX in the actual context sensing device,
and generates a signaling notification upon any
event that takes place at the PBX for a given
extension. An incoming call, outgoing call, call
end, call cancel, and so on will generate a notification from the PBX to an external monitoring
application.
In the presented design that event notification is carried over a computer telephony interface (CTI) protocol supported by the PBX,
which is oriented to report specific telephony
events. Although there are other choices, the
option selected is the Computer Supported
Telecommunications Applications (CSTA) PhaseIII protocol, specified by the European Computer Manufacturers Association (ECMA) [9],
supported by (or on the roadmap of) many PBX
manufacturers. This protocol consists of a complete call control signaling specification, imple-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
mented through a set of XML commands whose
format is defined by ECMA, carried on two possible transport options, SIP or Simple Object
Access Protocol (SOAP)/HTTP. In addition, the
PBX may also receive telephony commands
from external computing functions over CSTA
PhaseIII (release an ongoing call, reject an
incoming call, establish a call to a specific destination, etc.). The CSTA PhaseIII XML message
sent to the PBX to clear a specific connection (a
voice call) is the following.
<?xml version=”1.0” encoding=”UTF-8”?>
<ClearConnection xmlns=”http://www.
ecma-international.org/standards/ecma323/csta/ed3”>
<connectionToBeCleared>
<callID>123456789</callID>
<deviceID> tel:+34913374005 </deviceID>
</connectionToBeCleared>
</ClearConnection>
The CSTA PhaseIII events generated from
the PBX to an external entity are information
about the context of the user.
The procedure to capture similar context
information at the mobile network side is the
presence and registration status management of
the mobile cellular networks, a capability already
available in commercial mobile networks. Such
capability can provide updated information
about the location, registration status, and availability of cellular devices. In our mobile network
the location and registration information is
stored at the home location register (HLR) of
the home network of the mobile device, and is
updated during registration procedures via SS7
signaling protocols. The registration information
is sent from the mobile network to the convergent service layer via a SIP interface, through a
protocol gateway deployed at the mobile network.
The objective is to route all such contextual
information to a context enabler located at the
convergent service layer, accessible from both
fixed and mobile service layers. However, the
development of a context enabler platform was
not considered within the timeframe of the project due to the availability of existing elements
already deployed in the operator service layer
which can cover that need. The simplicity of the
context model to be followed (presented below)
made it clear that having a dedicated platform
for context management was suboptimum given
the time to market objectives. The development
of a complete context management enabler
became a parallel working direction with different time constraints.
The existing platform where the available
context of the user will be located is the OMA
presence server. Context information will be
routed to the presence server, and the context of
the user will be stored as a parallel presence status.
IMPLEMENTATION PHASE
A very important feature of the system on the
fixed side is the wide geographical distribution of
the PBXs. There is a PBX in each customer
premises (enterprises with several distributed
premises), meaning a big number of PBXs in a
wide geographical area (1000+ PBXs deployed
in several cities). There may also be enterprise
customers with a much smaller geographical distribution (or no distribution at all).
That is a critical aspect for the transport network. Initially, global connectivity over a SIP
overlay was considered. However, given the geographical distribution of the network elements
and the different network domains the signaling
needs to traverse (in some cases operated by the
enterprise customer itself), a transport protocol
with better interoperability and connectivity
characteristics was finally considered. Based on
this, the CSTA PhaseIII transport option chosen
is HTTP/SOAP. Given that after preliminary
laboratory testing the estimated link bandwidth
consumption imposed by this signaling load is
minimum (on the order of hundreds of kilobits
per second), the best option available was to use
the data link purchased by enterprise customers
at each premises to get external data connection.
In those premises where the external data access
is heavy loaded, additional digital subscriber line
(xDSL) access may be deployed (possible but
unlikely).
Another key aspect that affected the implementation even more than the geographical distribution is the diversity of PBX manufacturers.
There will be several PBX manufacturers
involved (initially two manufacturers were
involved, but additional ones may be included in
order to offer these services to all enterprise customers). Getting all manufacturers involved to
implement a coherent CSTA PhaseIII protocol
with the same behavior of PBXs in a robust way
was an unrealistic approach. PBXs from different manufacturers would have different implementations of CSTA PhaseIII, which would
make it impossible to get all the systems working
properly with tight time-to-market constraints.
Different manufacturers would mean different
CSTA PhaseIII flavors. This implementation
issue affected the architecture itself, calling for
the definition of an additional element, the
OBG, whose mission is to solve the inter-PBX
vendor compatibility problem. The function of
the OBG is to act as a single entry point of contextual signaling traffic from all the PBXs. All
the particularities of different PBX manufacturers would be considered at that element, and the
OBG would perform adaptations from the different CSTA PhaseIII flavors from different
PBX manufacturers into a CSTA PhaseIII implementation that is standard and stable within the
operator service layer domain. So the OBG
would deal with the interoperability aspects
among the different interface implementations
from the variety of PBX manufacturers.
The drawback of including this new element
in the architecture is mainly the significant
increase in the development costs of the project.
However, such protocol adaptation in the end
decreased the operational costs significantly. So
it was proven to be the right decision also in
terms of global costs. This is one of the main
project conclusions presented below.
Finally, it is worth clarifying that, since the
OBG has been defined as a single entry point
for CSTA PhaseIII signaling from the PBXs, it is
IEEE
BEMaGS
F
Different manufacturers would mean
different CSTA
PhaseIII flavors.
This implementation
issue affected the
architecture itself,
calling for the
definition of an
additional element,
namely the OBG,
whose mission is to
solve the inter-PBX
vendor compatibility
problem.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
37
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
The delay observed
for a single CSTA
PhaseIII signaling
transaction from PBX
to the hosted service
is in the order of
100 ms. For a
complete scenario in
which a call is routed
and the presence
status is updated,
the whole delay is in
the order of 500 ms.
a network element located at the operator service layer. Accordingly, the interface between
the OBG and the context enabler, as well as with
any other service, will be an operator network
protocol. Given that this is a convergent architecture based on an IMS core network, it was
quickly considered to implement such an interface over a SIP transport protocol. This would
make any later integration between the OBG
and additional services much easier.
Since SIP is chosen for the interface between
the OBG and the context enabler, the CSTA
PhaseIII messages received from the PBXs over
HTTP/SOAP will be parsed by the OBG and
adapted to standard CSTA PhaseIII over SIP
(uaCSTA), screening any particularity of the
PBX manufacturer implementation. The final
XML CSTA PhaseIII message is then embedded
in the body of a SIP INFO method, as stated by
the uaCSTA PhaseIII specification [10]. The
CSTA PhaseIII XML message is the same
regardless of the transport protocol (SIP or
HTTP/SOAP).
INTEROPERABILITY TESTING
Both the selection of SOAP as the protocol for the
CSTA PhaseIII message transfers and the inclusion of OBG in the architecture respond to the
objective of improving interoperability with minimum development. So interoperability testing was
an important phase before commercial rollout.
Again, to optimize time to market, an agile
development mechanism [11] is followed. Short
development cycles are implemented (1–2 mo),
and different levels of interoperability testing were
performed at the end of each development cycle.
PBX Level — A project structure based on
sequential functional iterations is followed jointly by the PBX manufacturers and the operator.
Periodically, the manufacturer releases a set of
functionality, and signaling compliance is validated by the operator via a simulator [12] that can
be quickly adapted to the special implementation of each PBX manufacturer.
OBG Level — The OBG element also needs to be
validated. The same procedure as with the PBX
manufacturers is followed. The engineering simulation tool is also used in this interoperability testing.
End-to-End Level — Finally, all elements are
installed in a laboratory environment with real
phones and fixed PBX extensions. Some interoperability issues are still identified, but this phase
is mostly a functional validation by the operator.
THE SERVICE USE CASE DEPLOYED:
CONTEXT-AWARE
CALL ROUTING SERVICE
There is currently no known (as far as we are
concerned) commercially deployed convergent
service that is able to route incoming calls to
corporate lines (PC, mobile or a fixed deskphone) to the appropriate destination based on
a specific situation (e.g., context) of the employees associated with the corporate lines.
38
Communications
IEEE
A
BEMaGS
F
The flexibility of the architecture presented in
the previous sections is demonstrated with the
intelligent routing service presented in this section. The basic use case of the service is presented in Fig. 3, in which an incoming call arrives to
the PBX targeting an employee subscribed to
the intelligent routing service [13].
In Fig. 3, when a call arrives to the PBX from
the PSTN, the PBX temporarily stops the call
progress and sends a CSTA PhaseIII event to
the OBG, requesting a route for the call
(CSTA_RouteRequest). The trigger to the
hosted service is executed by the PBX before a
call is routed to the destination extension, taking
into account that incoming calls can also originate from another internal extension of the
PBX.
The OBG performs the protocol adaptations
required based on the PBX manufacturer and
sends the request for a route to the intelligent
routing service by including the same
RouteRequest XML over a SIP INFO method.
The intelligent routing service has a specific
business logic implemented to select the destination of the call. That service logic includes, at a
high level, checking a list of prioritized destinations and the situation of each one by querying
the IMS presence server or additional information repositories like corporate calendar services.
If the destination user is online at any of his/her
phone devices, the service selects that destination and responds to the OBG with a command
(CSTA_RouteSelect) including the destination for the pending call.
Again, the OBG performs protocol adaptations as required, and sends to the PBX the
selected route for the call that is waiting to be
delivered. This selected destination may be a
fixed extension (connected to the PBX or to
another PBX) or a mobile extension. If it is a
mobile destination, the PBX will then need to
forward the call through the public mobile network. The selected destination might even be a
PC-based softphone that the PBX can reach
through the corporate private network.
Once the call is successfully established, the
context enabler is updated with the new user
context, again via CSTA PhaseIII notification.
Similar services can be designed, using the
contextual information stored at the presence
server (or a dedicated context enabler platform),
making use of the architectural mechanisms
deployed.
In terms of performance figures, the delay
observed for a single CSTA PhaseIII signaling
transaction from PBX to the hosted service is on
the order of 100 ms. For a complete scenario in
which a call is routed and the presence status is
updated, the whole delay is on the order of 500 ms.
Both values are consistent with the figures observed
in a laboratory environment, with very small variance values. However, much stronger performance
testing is required in a later deployment phase, as
the current number of customers is still growing.
LESSONS LEARNED
During the design of the architecture and the
deployment, several lessons can be identified to
consider in future developments with similar
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Public
network
PBX
OBG
Intelligent
routing service
IMS
Presence
server
F
Due to the limited
complexity of the
first service that will
INCOMING CALL
use the contextual
Stop call
information stored at
[WSCSTA]: Incoming call Event
the Presence Server,
the number of
[uaCSTA]: Incoming call Event
possible context
[uaCSTA]: Incoming call Event
statuses is small.
The range of values
Contextual
service
logic
[SIP]: SUBSCRIBE
might be increased
[SIP]: NOTIFY
for other contextual
services with little
effort.
[uaCSTA]: Selected Destination
[uaCSTA]: Selected Destination
[WSCSTA]: Selected Destination
Call
routing
[WSCSTA]: New Context Status
[uaCSTA]: New Status
[uaCSTA]: New Status
Context
update
Figure 3. Signaling flow for the intelligent routing service, based on the contextual architecture presented.
purposes.
The capabilities and protocols supported by the
different PBX manufacturers are very diverse.
That happens in any context-aware system
deployed by telco operators. The types and sources
of user context are also very diverse, and interoperability and protocol adaptations are required.
During the analysis phase, it was clear that
the number of protocols used within the operator network should be minimized to avoid interoperability issues. Since the control layer at the
operator network is IMS, the protocol used within both that domain and the service layer is SIP.
This means a simplification of the transport network configuration at the operator network. If a
different/additional protocol is to be used, it
would require modifications of the traffic flows
and verifications of connectivity across existing
firewalls. From an operational perspective that
would mean huge problems.
The transport protocol chosen to carry the
information from PBX to the operator network
is SOAP/HTTP. This option is selected as
opposed to SIP because of the simplicity of the
initial deployment and subsequent upgrades or
maintenance tasks. The data networks deployed
to connect customer premises with the operator
are not usually prepared to support SIP without
significant configuration updates.
Due to the limited complexity of the first service that will use the contextual information
stored at the presence server, the number of
possible context statuses is small. The range of
values might be increased for other contextual
services with little effort.
An additional element to develop was a single entry point into the operator network that
provides signaling adaptation into a secure and
standard protocol. The benefits for the testing
phase, the isolation of PBX vendor-specific
issues, and the subsequent securization of operator service layer generated project benefits that
justified the associated development cost.
CONCLUSIONS AND FUTURE WORK
In this work an architecture to handle the context information of telco users is presented. This
information is sent from the different access networks to a context server (enabler) hosted at the
operator service layer. Due to time to market
reasons, the context management server is collocated with the OMA presence server.
This proposed architecture is designed to be
future-proof and flexible to allow other services
to take advantage of the user context informa-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
39
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
This proposed
architecture is
designed to be
future-proof, and
flexible to allow
other services to take
advantage of the
user context
information,
becoming with small
development costs
context-aware
tion, becoming with small development costs
context-aware services. In order to do that, the
signaling protocols within the operator service
layer are properly selected.
A context-aware convergent service is
designed with this architecture. The objective of
the service is intelligent context-aware incoming
call routing for a corporate telephony network
(both fixed and mobile).
The future lines of work are twofold:
• Design and integrate new services into this
architecture, enhancing the added value of
the operator commercial portfolio.
• Develop a dedicated context management
platform where specific and more complex
information processing algorithms may be
implemented, including context semantic
processing and user profiling, among others.
services.
REFERENCES
[1] D. Saha and A. Mukherjee, “Pervasive Computing: A
Paradigm for the 21st Century,” Computer, vol. 36, no.
3, Mar. 2003, pp. 25–31.
[2] J. Sun, Z.-H. Wu, and G. Pan, “Context-aware Smart
Car: From Model to Prototype,” J. Zhejiang Univ. — Science A, vol. 10, no. 7, July 2009, pp. 1049–59.
[3] M. Baldauf, S. Dustdar, and F. Rosenberg, “A Survey on
Context-aware Systems,” Int’l. J. Ad Hoc Ubiquitous
Comp., June 2007, pp. 263–77.
[4] 3GPP TS 23.228 “IP Multimedia Subsystem (IMS): Stage
2”; http://www.3gpp.org
[5] H. van Kranenburg et al., “A Context Management
Framework for Supporting Context-aware Distributed
Applications,” IEEE Commun. Mag., vol. 44, no. 8, Aug.
2006, pp. 67–74.
[6] C. Chi et al., “IMS Presence Server: Traffic Analysis and
Performance Modeling,” IEEE Int’l. Conf. Net. Protocols,
Oct. 19–22, 2008, pp. 63–72.
[7] M. Schmidt et al., “IMS Interoperability and Conformance Aspects,” IEEE Commun. Mag., vol. 45, no. 3,
Mar. 2007, pp. 138–42.
[8] S. Arbanowski et al., “I-centric Communications: Personalization, Ambient Awareness, and Adaptability for
40
Communications
IEEE
A
BEMaGS
F
Future Mobile Services,” IEEE Commun. Mag., vol. 42,
no. 9, Sept. 2004, pp. 63–69.
[9] ECMA Standard 348 “Web Services Description Language (WSDL) for CSTA Phase III,” June 2009.
[10] ECMA Tech. Rep. TR/87, “Using CSTA for SIP Phone
User Agents (uaCSTA),” June 2004.
[11] M. Fowler and J. Highsmith, “The Agile Manifesto,”
Aug. 2001; http://agilemanifesto.org/
[12] J. M. González, A. Cadenas, and O. Solá, “Adaptation
Middleware to Enable Presence and Call Control for
Corporate Fixed Lines: Evolution to Convergent Network over IMS,” NGNM ‘08, Sept. 2008.
[13] A. Cadenas et al., “Distributed PBX Gateways to
Enable the Hosted Enterprise Services Architecture in a
NGN Scenario,” 1st ITU-T Kaleidoscope Academic Conf.,
May 12–13, 2008, pp. 203–10.
BIOGRAPHIES
ALEJANDRO CADENAS [M] (cadenas@tid.es)
_________ is currently a project manager in Telefónica Research and Development,
focusing on IMS, NGN, and innovation in end-user services,
network evolution of the operator infrastructure toward
user-centric services, context-aware hosted services, and
protocol adaptations. Previously he was a senior systems
engineer in Motorola Inc. for several years workingon network design and IMS control layer design. He is a Ph.D.
candidate researching convergent contex-aware architectures and interoperation with telco services.
ANTONIO SANCHEZ-ESGUEVILLAS [SM] (a.sanchez-esguevillas@
_____________
ieee.org) has managed innovation at Telefónica (both corporate and R&D), Spain. He is also an adjunct professor at
the University of Valladolid. His research interests relate to
services and applications. He is an Editorial Board member
of IEEE Communications Magazine and IEEE Network,
founder and Chairman of the IEEE Technology Management Council Chapter Spain, guest editor of IEEE Wireless
Communications, IEEE Communications Magazine, and IEEE
Network, and has served on the TPCs of ICC, GLOBECOM,
PIMRC, WCNC, HealthCom, CCNC, and VTC.
BELÉN CARRO (belcar@tel.uva.es)
__________ is an associate professor at
the University of Valladolid, where she is director of the
Communication and Information Technologies (CIT) laboratory. Her research interests are in the areas of service engineering, IP broadband communications, NGN and voice
over IP, and quality of service. She has extensive research
publications experience as author, reviewer, and editor.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
__________________________________________
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
The Design and Implementation of
Architectural Components for the
Integration of the
IP Multimedia Subsystem and
Wireless Sensor Networks
May El Barachi, University of Quebec
Arif Kadiwal, Nuance Communications
Roch Glitho, University of Quebec and Concordia University
Ferhat Khendek, Concordia University
Rachida Dssouli, Concordia University and United Arab Emirates University
This article is an extended
version of the article presented at IEEE VTC
2009-Spring under the
title of “The Design and
Implementation of a
Gateway for IP Multimedia Subsystem/Wireless
Sensor Networks Interworking.”
42
Communications
IEEE
ABSTRACT
INTRODUCTION
The IP multimedia subsystem is becoming the
de facto standard for IP-based multimedia services, while wireless sensor networks are gaining
in popularity due to their ability to capture a
rich set of contextual information. Integrating
the sensing capabilities of WSNs in the IMS can
open the door to a wide range of context-aware
applications in areas such as wireless healthcare
and pervasive gaming. We have previously proposed a presence-based architecture for
WSN/IMS integration. This architecture relies
on two key components: a WSN/IMS gateway
acting as an interworking unit between WSNs
and the IMS; and an extended presence server
serving as a context information management
node in the core network. In this article we focus
on the design and implementation of these two
components. Furthermore, two applications (a
pervasive game and a personalized call control
application) are used to concretely show how
new applications can be developed using our
architecture. Performance has also been evaluated. Several important findings were made in the
course of this work; one is that the IMS integration with a large and evolving variety of WSNs
may be a never-ending endeavor — the gateway
requiring constant upgrading due to the lack of
standard APIs for the interaction with sensors
produced by different vendors. Another finding
is that while the introduction of context as an
application building block in the IMS ensures
the availability of additional contextual information in the network and enables fast and easy
development of context-aware applications, the
lack of mature IMS application development
toolkits remains a roadblock.
The Third Generation Partnership Project
(3GPP)-defined IP multimedia subsystem (IMS)
is becoming the de facto standard for IP-based
multimedia services [1]. It consists of an overlay
control and service layer that is deployed on top
of IP-based mobile and fixed networks in order
to enable the seamless provisioning of IP multimedia services to end users. Wireless sensor networks (WSNs) are formed by a set of distributed
sensor nodes that collaborate to monitor physical, environmental, and physiological conditions
[2]. Such networks are increasingly popular
because they can capture a rich set of contextual
information that can be used for a wide range of
applications.
Context awareness is the ability to use contextual information to provide relevant information and/or services to users. By integrating the
sensing capabilities of WSNs in the IMS, a rich
set of contextual information can be exploited to
provide new and personalized multimedia services to IMS users. Examples of such services
include wireless healthcare applications monitoring and interpreting patients’ physiological data
and offering them personalized medical assistance for problematic health conditions; pervasive games involving interaction with
physical/virtual objects and characters, and using
the game context to adapt the players experience; and lifestyle assistance applications making
use of users’ situational information to assist
them in their daily activities (e.g., training and
shopping).
We have previously proposed a presencebased architecture for WSN/IMS integration.
This article focuses on the design and implemen-
0163-6804/10/$25.00 © 2010 IEEE
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
WSNs
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
WSN/IMS
Gateway
(PEA)
Proprietary
sensors
interfaces
Pexb
(Internal)
defined in the 3GPP
architecture. One of
Presentity presence proxy
ISC
P-CSCF
F
types of PAs are
IMS core
entities (e.g.
CSCFs)
Pwp
Pexa
BEMaGS
We note that several
Pwn
Extended
PS (AS)
A
Watcher presence proxy
these types is the
Presence External
S-CSCF
I-CSCF
S-CSCF
Agent (PEA)
P-CSCF
responsible of
Pw =
Mw
Pw =
ISC
Pw =
Gm
publishing
information provided
HSS
IMS AS
(E.g. GS)
IMS user
applications
by external
entities/networks
about the user.
Figure 1. The WSN/IMS integrated architecture.
tation of the key components of our architecture, and illustrates how applications (leveraging
its capabilities) can be built, using two concrete
examples. Furthermore, the article presents the
results of the performance evaluation we conducted using one of the developed applications,
and reports on what we learned from this project.
The next section gives an overview of the previously proposed architecture. In the following
section, the architectural components’ design is
presented. This is followed by a description of
the prototype implementation and proof-of-concept applications. We then present the performance evaluation. The last section ends the
article with a discussion of related work and the
lessons learned.
AN OVERVIEW OF THE WSN/IMS
INTEGRATED ARCHITECTURE
Figure 1 depicts the architecture we proposed in
[3] to enable the integration of WSNs in the
IMS.
This architecture is an extension of the 3GPP
presence architecture, which focuses on the
management and dissemination of user presence
information (a subset of context information)
within the network. The 3GPP presence architecture relies on five main functional entities:
presence agents (PAs), which make information
available to the network in a standard format
and via standard interfaces; the presence server
(PS), responsible for the management of information published by agents; the presence list
server, responsible for group list management;
presence proxies, acting as inbound/outbound
proxies to the presence network by performing
routing, security, and charging functions, and
whose roles are assumed by call session control
functions (CSCFs); and watchers, acting as information consumers. We note that several types of
PAs are defined in the 3GPP architecture. One
of these types is the presence external agent
(PEA), responsible for publishing information
provided by external entities/networks about the
user.
To achieve WSN/IMS integration, we
assigned the role of PEA to the WSN/IMS gateway, which will publish information captured by
WSNs (after proper processing/formatting) to an
extended presence server (capable of managing
the different types of information provided) via
a trusted node (a presence proxy) over the Pexa
interface. In addition to the indirect interactions
over the Pexa interface, the WSN/IMS gateway
directly interacts with the PS over the Pexb interface for the management of subscription policies
(enabling information access control). We note
that the WSN/IMS gateway can be considered
specialized user equipment, only used for the
management and publication of sensory information in the 3G network.
Other entities such as IMS application servers
(e.g., game servers) and IMS user applications
can access the information managed by the PS
via presence proxies using the Pw interface (corresponding to the IMS service control [ISC] and
Gm interfaces), while IMS core network entities
acting as watchers (e.g., CSCFs) can directly
access this information using the newly defined
Pwn intra-operator interface (without triggering
the generation of charging records). It should be
noted that the XML Configuration Access Protocol (XCAP) [4] is used over the Pexb interface,
while an optimized version of the SIMPLE protocol [5] is used over the Pex a , Pw, and Pw n
interfaces.
To enable the management and dissemination of the collected sensory information, we
proposed an extension of the standard presence
information model (i.e., the presence information data format [PIDF]) in order to allow the
representation of the additional types of information captured by WSNs (i.e., spatial, physiological, and environmental data) in a standard
format, while allowing the distinction between
the different types of entities to which the information relates, as follows: To allow the encapsulation of physiological and environmental data
within a presence document, two new subelements (physiologicalData and environmentalData) were added to the existing tuple
element. Each of these subelements was further
divided into other subelements. For instance, the
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
43
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
The abstraction layer
is responsible for
Abstraction layer
Info management functions
conveying the
Info rep.
information captured
by WSNs to the IMS,
after the proper
Info
acquisition
module
Processed
WSN info
Data model rep.
XML
formatter
- PIDF
- Mapping
table
processing and
formatting. This layer
Policies
rep.
consists of two types
of functions
(i.e., information
Extended PEA
management
functions and
Support functions
Trigger
handler
Publisher
Events
monitor
Registration
and security
ISIM
app.
Capabilities
publication
- Subscription
authorization
policies
- Publication
policies
Information
access control
support functions).
WSN interface
3G interface
Connectivity layer
Figure 2. The WSN/IMS gateway architecture.
complex element environmentalData is
divided
into
ambientTemperature,
soundLevel, lightIntensity, and relativeHumidity subelements, in addition to a
qualityInfo subelement and any number of
extension subelements.
For spatial information we leveraged the
existing GEOPRIV extension data element (specified in RFC 5491) for its representation, and
extended one of its child elements (civicLoc)
with refined location information such as room
ID, displacement direction, and relative distance
to other. Finally, to enable the distinction
between different types of entities (i.e., person,
object, place, or network) to which the information may relate, we added two new mandatory
attributes, entityType and entityDescription, to the existing presence element.
In our architecture, three information publication models were employed to enable flexible
and resource-efficient information exchange
within the network: interval-based publications,
in which information is published at regular time
intervals; event-based publications, in which
information is published when certain events are
detected; and trigger-based publications, in
which information is published upon receipt of a
publication trigger from the PS.
ARCHITECTURAL
COMPONENTS’ DESIGN
As shown in the previous section, our architecture relies on two main components in its operation: the WSN/IMS gateway and the extended
presence server. In the subsequent subsections,
we describe the designs of these two components.
44
Communications
IEEE
THE WSN/IMS GATEWAY ARCHITECTURE
The WSN/IMS gateway plays a key role in our
architecture by acting as an intermediary
between WSNs and the 3G network. Figure 2
depicts the proposed gateway architecture, which
consists of two layers: a connectivity layer and an
abstraction layer. The connectivity layer includes
a dual networking interface, ensuring connectivity to both WSNs and the 3G network. The
abstraction layer is responsible for conveying the
information captured by WSNs to the IMS, after
the proper processing and formatting. This layer
consists of two types of functions (i.e., information management functions and support functions) that are described in detail in the coming
subsections.
The Support Functions — The support functions are realized by the following modules: the
registration/security module, capability publication module, and information access control
module.
The registration/security module is the first
module invoked when the gateway (GW) is put
in service. It interacts with the ISIM application
(contained in the gateway’s SIM card) to get the
information required for IMS registration and
security association establishment (e.g.,
public/private identities and the long-term
secret), builds the first SIP REGISTER message,
and interacts with the capabilities publication
module that inserts the gateway capabilities
information (e.g., the type of information provided by the GW) in the message body. The registration module then carries out the rest of the
IMS registration procedure normally.
After the registration phase, the information
access control module communicates with the PS
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
to set the required subscription authorization
policies (which are preconfigured in the GW’s
policies repository). Those policies are used by
the PS to install filters that determine which
watchers are allowed to access the information
related to a certain contextual entity, thus preserving information privacy. In addition to subscription policies, the GW’s policies repository
also contains publication policies indicating the
types of information that should be published at
regular time intervals and the ones that are published based on events’ detection.
The Information Management Functions —
The information management functions are performed by a set of information acquisition modules, an XML formatter, an extended PEA, and
three repositories. The first repository (the data
model repository) contains the extended PIDF we
defined as an information model and mapping
tables correlating IMS entities’ IDs (identifying
the entities whose information is being published
in the IMS) to sensor IDs (identifying the sensors capturing information related to a certain
entity), while the second (the information repository) contains the processed WSN information
that is persistently stored in the GW for future
publications. The third repository is the policies
repository presented above.
The information acquisition modules are specialized components that enable the interaction
with various WSNs. There is one acquisition
module per WSN type. Such a module is capable
of extracting sensor-specific data from WSN
messages, and preprocessing this information (by
performing data fusion and consistency checking) before storing it in the GW’s information
repository. When the WSN comes with a middleware (e.g., the one proposed in [6]), the module interacts with the middleware instead of
interacting directly with the WSN. In this case
less or no preprocessing may be required,
depending on the middleware capabilities.
The extended PEA represents the heart of the
WSN GW. It publishes the WSNs’ information
to the IMS, based on the publication policies
defined in the GW. Two modes of publications
are supported by the extended PEA: the proactive mode, in which information is actively published by the GW at regular time intervals or
when certain events are detected; and the reactive mode, in which information is only published upon receipt of a trigger from the PS.
These two modes of publication are realized
by the PEA submodules as follows: Based on the
publication policies, the publisher saves a list of
information that should be proactively published
at regular time intervals; following those intervals, it consults the information repository to
retrieve the needed information, which is then
passed to the XML formatter (if the information
is not yet in the standard format). The XML formatter consults the IDs’ mapping tables and the
extended PIDF in order to represent the processed information in a standard format, and
then returns the resulting XML document to the
publisher, which publishes it to the PS. Similarly,
the events monitor saves a list of information to
be proactively published upon detection of
events (e.g., publish temperature when above
- Extended PIDF
- Contextual info
- Info access policies
XML/parser/
formatter
Publication and
Trigger
subscription
generator
manager
Notifier
Events
monitor
IEEE
BEMaGS
F
Info/policies
repository
Authentication
and authorization
module
Presence service logic
SIMPLE stack (SIP servlet API)
Figure 3. The extended presence server architecture.
30°C), and keeps interacting with the information repository to detect the occurrence of those
events. Once an event is detected, the events
monitor interacts with the publisher, which will
fetch the needed information and publish it
(after proper formatting) to the PS. As for the
trigger handler, it does not actively publish any
information. However, once it receives a publication trigger (from the PS), it contacts the publisher, which will convey the needed information
to the PS.
THE EXTENDED PRESENCE
SERVER ARCHITECTURE
Figure 3 depicts the software architecture of the
extended presence server serving as the context
management node in our system.
This architecture consists of protocol and service-related components. The protocol supported in this case is the SIMPLE protocol, which is
accessible via the SIP servlets application programming interface (API). The service component consists of a presence service logic module
implementing the logic of the PS engine. This
module relies on several submodules in its operation: a publication/subscription manager that
handles information publications and subscriptions from presence agents and information
watchers; a notifier that creates and sends information notification messages based on received
subscriptions (these notifications could be sent
following regular time intervals or upon the
detection of events); an events monitor that
monitors the collected information and detects
the occurrence of events that could possibly lead
to information notifications; a trigger generator
that generates publication triggers to prompt the
publication of information that is missing or not
fresh enough in the network; and an authentication and authorization module that is responsible for the authentication of publishers/
watchers and the enforcement of subscription
authorization policies. It should be noted that
these submodules rely on an XML parser/for-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
45
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Extended PS (SIP servlet)
SIP container
WSN/IMS
GW
Server side of
game (SIP servlet)
Server side of
SenseCall App.
(SIP servlet)
SIP / SIMPLE
Emulated CSCF
SIMPLE
Diameter
Emulated
HSS
SIP
Game client (deployed on
phone) and SenseCall client
(deployed on laptop)
UE
Figure 4. The prototype components.
matter for the extraction of information from
received messages and for XML formatting of
information to be inserted in newly created messages; and they rely on an information/policies
repository for the storage of the extended PIDF
(used as XML schema), the information access
policies, and the contextual information that is
persistently stored for future notifications.
PROTOTYPE IMPLEMENTATION AND
PROOF OF CONCEPT APPLICATIONS
To build a proof-of-concept prototype of our
architecture, we used Ericsson’s Service Development Studio (SDS) [7] as the implementation
platform and extended the existing JAIN presence server [8] to fit our design. The WSN/IMS
GW was implemented from scratch. Furthermore, to illustrate how new applications can be
developed using the capabilities of our system, a
pervasive game called Fruit Quest and a personalized call control application, Sense Call, were
developed. In this section we start by giving a
general description of these two applications,
and then discuss the prototype architecture
before presenting the setups used to test the
application scenarios.
PROOF OF CONCEPT APPLICATIONS
Fruit Quest is a strategic pervasive game
designed in our laboratory. In this game, players
are assigned plantation zones, in addition to
some virtual game objects (e.g., fruits, walls,
bombs, and virtual money). WSNs are used to
detect and convey the location of players and
their presence in zones to the network. Players
physically move between plantation zones within
the game area, and as they move, they see the
zones appearing on their terminals and get notifications about game events. They can also communicate with each other using IMS instant
messaging. When players are in their plantation
zones, they can plant fruits and add defensive
walls for protection. When in rivals’ zones, players can pick fruits and attack the zones using
bombs. When all defensive walls in a zone are
destroyed, the zone can be occupied by rivals.
When the time of the game ends, the player with
the highest number of zones and fruits wins the
game.
46
Communications
IEEE
A
BEMaGS
F
Sense Call [9] is a personalized call control
application previously developed in our laboratory as part of another project in which a webservice-based GW was used to enable
applications’ interaction with sensors. This application monitors users’ locations and enables the
automatic (pre-booked) establishment of a call
between two colleagues when they are in their
respective offices. To illustrate the capabilities of
our system, Sense Call was remodeled and
deployed in our WSN/IMS integrated environment.
PROTOTYPE ARCHITECTURE
The SDS is an Eclipse-based design and execution environment in which IMS applications can
be designed, deployed, and tested. One of the
features provided by SDS is an IMS simulated
environment simulating CSCFs, a home subscriber server (HSS), and an application server
acting as a container for the deployment of SIP
servlet-based services. Figure 4 illustrates the
different prototype components developed using
the SDS platform.
In the prototype the JAIN PS [8] (originally
relying on a JAIN SIP stack for communication)
was remodeled as a SIP servlet to enable its
deployment in the SDS application server. Furthermore, the presence server’s XML schema
was extended with the additional data elements,
and its logic was enhanced with the publication
trigger mechanism. The server side of the gaming application was implemented as a SIP servlet
and deployed in the SDS application server,
while the game clients were developed using the
SDS IMS client platform and installed on P990
Sony Ericsson phones. The server side of Sense
Call [9] (originally developed as a SIP-based
standalone Java application) was remodeled as a
SIP servlet and ported to SDS. Furthermore, the
application logic was modified to communicate
with the PS (instead of direct communication
with the GW) to obtain the required information. In this prototype two types of sensors were
used: MIT Cricket location sensors [10], accessible via the Cricket API; and the MTS300/Mica2
environmental sensor [11], accessible via the
Crossbow API.
As for the WSN/IMS GW, it was implemented as a Java-based extended presence agent relying on a Microsoft access database and a set of
APIs (the JAIN SIP, Cricket, and Crossbow
APIs) in its operation. All of the components of
the GW were implemented, except for the capability publication and access control modules,
which were omitted for simplicity.
PROTOTYPE SETUPS AND TEST SCENARIOS
As shown in Fig. 5a, the Fruit Quest game setup
consisted of two laptops and three phones, forming a WLAN, in addition to a set of MIT Cricket
location sensors [10]. The game clients ran on
the phones, while one of the laptops represented
the IMS simulated environment (including the
game server) and the other laptop represented
the WSN GW. The following interactions related to the pervasive gaming scenario were successfully tested: First, the WSN GW was
registered in the IMS. Then three players, each
carrying a phone with an attached Cricket listen-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
er, started moving between three game zones.
Cricket beacons mounted to the ceiling were
used in conjunction with the Cricket listeners
attached to the phones to determine the location
of the players. This information was conveyed
(by Cricket software running on phones) to the
WSN GW using TCP/IP communication. The
GW monitored the information received, and
when it determined that a player had moved to
another game zone, it published this event (using
a SIP PUBLISH message) to the extended PS,
which notified (using a SIP NOTIFY message)
the game server. This last then sent the appropriate game updates (based on the received
information) to the game clients hosted by the
players’ UE, which updated the game display.
A similar setup was used for the Sense Call
application, as depicted in Fig. 5b. In this case,
the application clients were installed on two laptops, while the IMS simulated environment hosted the server side of the application. The
following interactions were then successfully
tested: First, the two clients were registered as
IMS users, and the server side of the application
was used to schedule a call between them. Next,
the users, carrying their laptops (with attached
Cricket listeners), started moving throughout the
office space, and their location information was
conveyed to the GW. Upon detection of a location change, the GW published this event (using
a SIP PUBLISH message) to the PS, which notified the server side of the application (using a
SIP NOTIFY message). When the application
detected that two users were in their respective
offices, it established a third-party controlled call
between them by sending a SIP REFER message
to one of the users’ UE. This last accepted the
referral using a SIP 202 message, then sent a SIP
INVITE message to UE2. When the call was
established successfully, UE1 notified the Sense
Call application about the result of the referral
event.
PERFORMANCE EVALUATION
To evaluate the performance of our system, we
used the Fruit Quest prototype to collect some
measurements, focusing on the publication interactions (between the GW and the PS) and notification interactions (between the PS and the
game server). Spatial (i.e., location) and environmental (i.e., light/temp) data was collected, and
two performance metrics were used: the
response time (in milliseconds) and the network
load (in bytes).
In addition to the sensors, the testbed consisted of the following: the GW, running on a
Pentium 4/2.5 GHz laptop, with 512 Mbytes
RAM and Windows XP. This laptop was connected with an MIB510 sink node, allowing it to
communicate and collect data from sensor nodes
— this data being monitored using a MoteView
application installed on the laptop. A second
laptop with a similar configuration (i.e., 1.6 GHz
Intel Pentium Duo with 1 Gbyte RAM, running
Windows XP) hosted the IMS simulated environment (the CSCFs, HSS, and extended PS),
while a third laptop with an identical configuration hosted a second instance of the IMS environment in which the Fruit Quest game server
WSN/IMS
gateway
2. Event
detected:
game zone
change
IEEE
BEMaGS
F
4. NOTIFY
(game_zone_
change event)
3. PUBLISH (game_zone_
change event) / 200 OK
IMS simulated environment (CSCFs,
HSS, extended PS, game server)
Area 1
5. Game updates
Area 3
Area 2
1. Update
location
1. Update
location
User’s phone
(with attached
cricket listener)
1. Update
location
Cricket
beacons
User’s phone
(with attached
cricket listener)
User’s phone
(with attached
cricket listener)
(a)
WSN/IMS
gateway
2. Event
detected:
location
change
4. NOTIFY
(location_
change event)
3. PUBLISH (location_
change event) / 200 OK
5. Event detected
two users in their
offices
IMS simulated environment (CSCFs, HSS,
extended PS, SenseCall application server)
1. Update
location
1. Update
location
7. INVITE/
OK/ACK
6. REFER (refer_to: UE2;
method=invite)/
8. NOTIFY (refer_event)
User’s laptop
(with attached
cricket listener)
Room A
Cricket
beacons
User’s laptop
(with attached
Room B cricket listener)
(b)
Figure 5. Prototype setups: a) Fruit Quest game setup; b) Sense Call application
setup.
was deployed. It should be noted that the game
server’s logic was slightly modified (for testing
purposes) to subscribe/accept environmental
information from the PS, in addition to the location information it originally used. Furthermore,
three Sony Ericsson P990 phones, with attached
Cricket listeners and running the Symbian operating system and the IMS client platform (provided with SDS), hosted the game clients. Table
1 shows some of the measurements collected
using this testbed. These values are average
measurements over 20 trials.
In the measurements displayed in Table 1,
the response time for proactive publications is
calculated at the GW level as the time duration
between the moment when information is
accessed by the GW’s publisher module (from
the information repository) and the message is
created/sent, until a successful publication
response is received from the PS. For reactive
publications, the response time (also measured
at the GW level) is calculated from the moment
a publication trigger (i.e., a SIP OPTIONS mes-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
47
A
BEMaGS
F
Communications
IEEE
Operation
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Scenario
Response time
(ms)
Network load
(bytes)
Proactive-location
information
205
1139
Proactive-environmental information
178
1067
Reactive-location
information
228
2371
Reactive-environmental information
214
2300
Location
information
224
1335
Environmental
information
164
Publication
Notification
1241
Table 1. Network load and response time measurements.
sage) is received from the PS, acknowledged and
responded to by a PUBLISH message, until a
successful publication response is returned by
the PS. As for notifications, the response time is
measured at the PS, from the moment the information is internally accessed and the message is
created/sent, until a successful notification
response is received from the game server.
Several types of comparative analysis were
made by examining the collected measurements,
two of which are presented here. The first analysis was made by comparing the performance of
the two modes of publication for the same type
of information (e.g., proactive location vs. reactive location), in order to calculate the overhead
introduced in the case of the reactive mode. This
overhead is caused by the exchange of an additional pair of SIP messages (i.e., OPTIONS and
OK messages) to trigger the publication, and by
the processing of the publication triggers contained in the OPTIONS message body. The average overhead, in terms of response time, ranges
between 23 ms (for location info) and 36 ms (for
environmental info) per operation, which can be
considered non-significant, since its effect will
barely be felt by the end user. The penalty in
terms of network load is nevertheless significant
(an increase of 1.2 kbytes/operation, for both
types of information). However, this penalty will
only be incurred occasionally since the reactive
mode is a secondary mode of operation only
used when the required contextual information
is not available (or is not fresh enough) in the
network.
By comparing the performance of one mode
of publication for two different types of information (e.g., proactive location vs. proactive environmental), we can see how the type of
information exchanged can affect the performance. The same type of comparison can be
made for notifications of different types of information. In general, we notice that the publications/notifications of environmental data achieve
better response times and induce lighter loads
48
Communications
IEEE
A
BEMaGS
F
than location-information-related interactions
(e.g., a decrease of 27 ms and 72 bytes for proactive-environmental publication in comparison to
proactive-location publication). This is because
the number of XML fields/tags required to represent location information is larger than what is
needed to model environmental data, thus
requiring more time for XML formatting and
generating bigger message payloads. The performance of location-information-related interactions could therefore be improved by using
another modeling schema that requires a smaller
number of tags for the representation of this
type of information.
RELATED WORK AND
LESSONS LEARNED
Several solutions have been proposed for the
integration of WSNs in the Internet, while few
others have investigated their integration with
3G networks. In this section, we discuss the solutions that are the most relevant to our work.
The e-SENSE architecture [12] aims at making ambient intelligence available to beyond 3G
networks to enhance their service provisioning
capabilities. This solution focuses on information
acquisition aspects by defining the protocol stack
to be implemented by sensor nodes as well as a
reference model for the WSN GW, but it does
not address issues related to information management in the core network. Moreover, the
proposed GW model is generic and does not
take into consideration IMS-specific requirements for WSN/3G integration.
TinyREST [13] and TinySIP [14] are solutions for WSN/Internet integration. Both solutions propose the use of application-level GWs
to enable the exchange of information between
WSNs and Internet clients. Although these solutions rely on standard IP protocols (i.e., HTTP
and SIP), they employ standalone GWs that are
built to be used directly by end users’ applications and cannot be integrated in the IMS to
leverage its other capabilities. The same limitation applies to the web-service-based GW [9]
previously developed in our laboratory.
During the course of this project, we learned
several important lessons. The first is that the
introduction of a sensor GW and an extended
PS in the IMS architecture enables the availability of additional contextual information in the
network. For instance, in our prototype, two
additional types of information were made available in the IMS: high-precision location information provided by Cricket sensors (e.g., the ID of
a room in a building) — information that constitutes a refinement of the location information
currently supported in mobile networks (mainly
low-precision cell ID information); and environmental data provided by the MTS300/Mica2
environmental sensor — information that is currently unavailable in a standalone IMS. Additional types of information could also be
supported by integrating other types of sensors,
such as the Zephyr BioHarness system providing
physiological and activity-related information.
Another lesson we learned while implementing the GW is that the development of systems
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
that interact with heterogeneous sensors coming
from different vendors is challenging. In our
case, our WSN/IMS GW interacted with location
and environmental sensors using two proprietary
interaction mechanisms (i.e., the MIT Cricket
API and the Crossbow platform), and had to
map their sensor-specific data into an abstract
format before transforming it into XML documents that are published using SIP in the 3G
network. Therefore, the interaction with additional types of sensors would require modifications/additions to the GW logic (mainly the
WSN interface and the information acquisition
module) each time a new sensor type is considered. This is due to the lack of a unified standard API enabling interaction with various
sensors. An initiative that may solve this issue is
the Zigbee standard [15], which aims at achieving interoperability between sensor nodes as well
as sensors and sinks, by defining a common protocol stack to be supported by all vendors. Currently, the Zigbee standard has specified the
protocol layers up to the transport level — the
application layer, however, remains to be
defined. We should also note that the modularity of our GW design helps to minimize this
impact by limiting to two the number of modules
that will be affected (by the change of the targeted sensor).
Another lesson related to the implementation
of the extended presence server is that there are
only a relatively small number of SIMPLE-based
open source presence servers. In this project we
chose the JAIN PS since it is easily extendible
and is based on the popular JAIN SIP stack with
which we were familiar from other projects.
However, like the other open source servers,
the JAIN server does not implement advanced
presence functionalities such as partial publications/notifications and event notification filters.
Furthermore, its design suffers from tight coupling between the presence server functionality
and the SIP proxy/registrar functionalities, which
are all bundled together. This resulted in duplicate functionality when this server was deployed
in SDS. For example, IMS users had to be registered twice, once to the CSCF and again to the
PS.
In relation to applications development, we
also learned that the introduction of context as
an application building block in the IMS facilitates the development of novel context-aware
value-added services. In fact, the functional separation between context management operations
and the logic of the applications (using the contextual information) enables the reuse of common mechanisms and concepts by different
applications, and abstracts developers from the
complexities of context acquisition/management
operations (e.g., the interaction with sensors
using proprietary APIs, and the processing of
the collected low-level information), thus speeding the application development process. This is
evidenced by the fact that the initial version of
the game, which was designed to directly communicate with sensors (in the initial phase of the
project), required close to two weeks for the
implementation of the logic related to sensory
information acquisition from the Cricket sensors,
with which we were not familiar. Additional time
would have been required if we had to learn how
to deal with other types of sensors. However,
porting this application to our WSN/IMS integrated environment, by replacing the original
information acquisition logic with standard SIMPLE-based interactions with the PS, only took us
one day — with the observation that minimal
effort would be required for the collection of
additional types of information from the PS,
since the same logic could be reused.
Finally, the availability of IMS development
toolkits is essential for the development and
testing of IMS-related prototypes and applications. However, the choices of freely available
toolkits are limited to Ericsson’s SDS and the
open source IMS core (OSIMS) [16] developed
by Focus. Unlike the OSIMS, which only offers
some of the IMS core nodes (i.e., CSCFs and
an HSS), SDS provides a more comprehensive
design and test environment by also offering an
integrated development environment, a set of
service APIs facilitating the development of
client/server side applications, an IMS terminal
emulator, an automated testing framework,
and a presence group management (PGM)
module.
Despite its attractive features, SDS does have
some limitations, the first being its lack of extensibility. In fact, since this toolkit is not open
source, it was not possible to extend its PGM
module, which obliged us to use an external PS
and remodel it according to the SIP servlet technology in order to deploy it in SDS. Furthermore, SDS did not support all the IMS
functionalities we needed; for instance, implicit
registration of identities was not supported by its
provisioning environment. Finally, it is important
to mention that the use of SDS requires basic
knowledge of the IMS operation and offers a
low level of abstraction in relation to its configuration, the provisioning of users, and the technology supported for the development of
server-side applications (i.e., the SIP servlet
technology). This is certainly a roadblock to
speedy application development, independent of
the type of applications involved.
IEEE
BEMaGS
F
It is important to
mention that the use
of SDS requires basic
knowledge of the
IMS operation and
offers a low level of
abstraction in
relation to its
configuration, the
provisioning of users,
and the technology
supported for the
development of
server-side
applications.
REFERENCES
[1] G. Camarillo and M. Garcia-Martin, The 3G IP Multimedia Subsystem, Wiley, Aug. 2004.
[2] I. Akyildiz et al., “Wireless Sensor Networks: A Survey,”
IEEE Commun. Mag., Aug. 2002.
[3] M. El Barachi et al., “A Presence-Based Architecture for
the Integration of the Sensing Capabilities of Wireless
Sensor Networks in the IP Multimedia Subsystem,” IEEE
WCNC ‘08, Las Vegas, NV, Mar. 2008.
[4] J. Rosenberg, “The XML Configuration Access Protocol
(XCAP),” IETF RFC 4825, May 2007.
[5] SIMPLE Working Group, “SIP for Instant Messaging and
Presence Leveraging Extensions (SIMPLE)”; http://www.
ietf.org/html.charters/simple-charter.html
[6] K. Aberer, M. Hauswirth, and A. Salehi, “A Middleware for
Fast and Flexible Sensor Network Deployment,” 32nd Int’l.
Conf. Very Large Databases, Seoul, Korea, 2006.
[7] “Ericsson Service Development Studio 3.1 — Technical
Product Description,” Feb. 2006; http://www.ericsson.
com/mobilityworld/developerszonedown/downloads/
docs/ims_poc/SDS_technical_description.pdf
_______________________
[8] JAIN-SIP-PRESENCE-PROXY; http://www-x.antd.nist.gov/
proj/iptel/nist-sip-downloads.html
__________________
[9] T. Ta et al., “Using Web Services for Bridging End User
Applications and Wireless Sensor Networks,” IEEE ISCC
‘06, Sardinia, Italy, June 2006.
[10] A. Smith et al., “Tracking Moving Devices with the
Cricket Location System,” MOBISYS, Boston, MA, 2004.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
49
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
IEEE
F
ROCH H. GLITHO [SM] holds a Ph.D. (Tekn.Dr.) in tele-informatics (Royal Institute of Technology, Stockholm, Sweden)
and M.Sc. degrees in business economics (University of
Grenoble, France), pure mathematics (University of Geneva,
Switzerland), and computer science (University of Geneva).
He works in Montreal, Canada, as associate professor at
ETS, University of Quebec, and as adjunct associate professor at CIISE, Concordia University. He has held several
senior technical positions (e.g., expert, senior specialist) at
Ericsson Canada and Ericsson Sweden. He has also served
as Editor-In-Chief of two IEEE Communications Society
magazines: IEEE Communications Magazine and IEEE Communications Surveys & Tutorials. (http://www.ece.concor_____________
dia.ca/~glitho/)
________
BIOGRAPHIES
FERHAT KHENDEK received a Bachelor’s degree in computer
engineering, option software, from the University of TiziOuzou, Algeria, and M.Sc. and Ph.D. degrees in computer
science from Université de Montréal. He is a professor with
the Electrical and Computer Engineering Department of
Concordia University. From 2001 to 2002, and from 2008
to 2009, he was a visiting researcher with Ericsson Research
Canada in Montreal. He has published more than 120 refereed research papers in journals and conference proceedings. His research interests are mainly in the design,
modeling, and analysis of real-time software systems, high
service availability, and value-added service engineering for
next-generation networks.
ARIF VALI KADIWAL received a B.S. degree in computer engineering from Sir Syed University of Engineering and Technology, Pakistan, in 2003 and an M.S. degree in electrical
and computer engineering from Concordia University in
2009. He worked at the Telecommunication Service Engineering Laboratory, a joint research laboratory between
Ericsson and Concordia University, from 2007 to 2008. His
work was concentrated in the area of the IP multimedia
subsystem and wireless sensor networks.
Communications
BEMaGS
[11] MTS300 Sensor; http://www.xbow.com/Products/
productsdetails.aspx?sid=75
_______________
[12] A. Gluhak et al., “e-SENSE Reference Model for Sensor
Networks in B3G Mobile Communication Systems,”
15th IST Summit, 2006.
[13] T. Luckenbach et al., “TinyREST: A Protocol for Integrating Sensor Networks into the Internet,” REALWSN
‘05, Sweden, 2005.
[14] S. Krishnamurthy, “TinySIP: Providing Seamless Access
to Sensor-based Services,” MOBIQUITOUS ‘06, San Jose,
CA, 2006.
[15] Zigbee Alliance, “Zigbee Specification 1.0,” June 2005;
http://www.zigbee.org/en/index.asp
[16] Open IMS Core; http://www.openimscore.org/
MAY EL BARACHI (elbar_m@ece.concordia.ca)
_______________ holds a Ph.D.
and a Master’s degree in electrical and computer engineering from Concordia University, Canada, and a Bachelor’s
degree in electronics and telecommunications engineering
from the Arab Academy for Science and Technology,
Egypt. She carried out her Master’s and doctoral research
as part of an industry/academia cooperation program
established between Ericsson Research Canada and Concordia University. She was also part of the IST Ambient Networks project — a European Union (EU) 6th framework
project. Presently, she is a postdoctoral fellow at the University of Quebec School of Superior Technology (ETS). Her
current research interests include service engineering, quality of service and adaptive resource management, context
awareness, virtual networks, and next-generation networks.
50
A
R ACHIDA D SSOULI received a Doctorat d’Université degree
in computer science from the Université Paul-Sabatier of
Toulouse, France, in 1981, and a Ph.D. degree in computer science in 1987 from the University of Montréal.
She is a professor with the Electrical and Computer Engineering Department of Concordia University and with the
College of Information Technology of the United Arab
Emirates University. She spent a sabbatical year
(1995–1996) at Nortel and was a visiting professor at
Abu Dhabi University, United Arab Emirates, from 2008
to 2009. Her research area is communication protocol
engineering, requirements engineering, and multimedia
applications.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Join
ComSoc
at 1/2 Year Rate!
www.ieee.org/join
Run Ahead of the Curve
IEEE Communications Society
Global Community of Communications Professionals
Membership benefits society members enjoy include:
IEEE Communications Magazine.
Wireless Communication Engineering
Technologies (WCET) Certification Program.
ComSoc e-News issues.
ComSoc Community Directory.
member-only products.
For free digital sample copy, go to
http://ww2.comsoc.org/Digital-Sample
____________ ______________
Headquarters - New York, USA
3 Park Avenue, 17th Floor
New York, NY 10016 USA
Tel: +1 212 705 8900
Fax: +1 212 705 8999
society@comsoc.org, www.comsoc.org
_________
Communications
IEEE
Singapore Office
Fanny Su Beh Noi, Manager
59E Science Park Drive
The Fleming, Singapore Science Park
Singapore 118244 SINGAPORE
Tel. +65 778 2873, Fax: +65 778 9723
f.su@ieee.org
______
China Office
Ning Hua, Chief Representative
Rm 1530, South Twr,
Raycom Info Tech Park C.
Haidian District
Beijing, 100190, China
Tel. +86 10 8286 2025, Fax: +86 10 8262 2135
n.hua@ieee.org
_______
IEEE Member Services
IEEE Operations Center
445 Hoes Lane,
Piscataway, NJ 08854-4141 USA
Tel: +1 800 678 IEEE (4333)
Outside USA & Canada: +1 732 981 0060
Fax: +1 732 562 6380
member-services@ieee.org, www.ieee.org
___________
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
Broadband Internet in EU Countries:
Limits to Growth
Ryszard Struzak, National Institute of Telecommunications
ABSTRACT
This article provides an analysis of broadband
Internet diffusion in 27 countries of the European Union. It proposes a simple model of its
growth and identifies the theoretical growth limits in each country. Some aspects of the European i2010 project implementation are
presented, discussed, and compared with the
model. Comments on bottlenecks and major barriers in the broadband Internet diffusion process
are also offered. The existing digital gaps are
irreducible in some cases and will exist as long as
the current development conditions continue.
The approach and analysis method proposed
here may be useful in examining limits of other
services or in other regions during the planning,
design, implementation, and performance tracking stages of existing or new services.
INTRODUCTION
Broadband Internet popularity is not uniform
and there are significant gaps or divides among
countries, regions, social groups, and so on. That
popularity is often expressed in terms of the
number of broadband access lines per 100 inhabitants, known as the penetration rate. The percentage of households with broadband Internet
access and percentage of enterprises with such
access are other metrics of the popularity.
These disproportions (digital gaps) constitute
an open issue for both developed and developing countries, and various projects have been initiated around the world to reduce them. The
European Union (EU), for instance, has
launched its i2010 project [1]. The EU is an economic and political union of 27 member states
offering the free movement of people, goods,
services, and capital. It represents about 500 million citizens and 22 percent of the gross world
product.
The EU countries differ significantly in size,
development degree, and wealth. For instance,
the gross domestic product purchasing power
parity per capita (GDP PPP) is $81,200 in Luxembourg and $12,900 in Bulgaria. Large differences also exist between geographical regions
and between social strata in some countries.
The aim of the i2010 project is to complete
the Single European Information Space. The
target is a broadband penetration rate of 90
52
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
percent (uniform across the whole EU) by the
year 2010 and higher percentages in later years
[2]. The project is well advanced and statistical
data indicate that every EU country has been
progressing fast. In spite of that success, the
gaps between countries still exist, and in some
cases they have increased during the project
lifetime.
Will the gaps ever vanish? As the raw statistical data do not directly answer this intriguing
question, this article proposes a simple approach
that leads to a reasonable answer. The article is
organized as follows. The next two sections present the approach used. A mathematical model
is proposed to infer intrinsic limits of the penetration rates in various countries and user
groups. Once the limits are known, it is straightforward to determine whether or not the gap
between two given countries or user groups will
vanish and to answer the question.
The limits are calculated for each of the 27
EU countries, and compared with the actual
penetration rate and the i2010 target. Some barriers to broadband diffusion are discussed.
Efforts required to reach the i2010 target are
evaluated and compared among countries to
infer possible bottlenecks. For that purpose, two
indices, the market index and effort index, are
proposed.
The article extends earlier work [3]. The presented approach is generic and may be applied
to other regions outside the EU and to other
services in the telecommunications space. It may
be useful when assessing the business viability of
a new or existing service, and when tracking the
business performance of a service after deployment.
MODELING
From time immemorial, people have wanted to
know the future. Earlier, they consulted oracles
and magic omens, but today they use mathematical models that are based on data, which represent the process under consideration. Having the
data, one selects an appropriate mathematical
function and matches it to the data. With such a
function, one can calculate (interpolate or
extrapolate) new data, outside the original set of
known (measured) data.
The model presented below is proposed not
to make specific predictions, but to explore and
understand better how the broadband Internet
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
penetration rate grows with time. That idea is
somewhat borrowed from The Limits to Growth,
a famous book by Donella H. Meadows and her
MIT colleagues, as the title of this writing indicates.
The growth of the Information Society and
Internet diffusion can be compared to a course
of virtual illness, where infected people become
Internet users. In case of real diseases we would
like to be all 100 percent immune; hence vaccinations and other preventive measures. With
Internet diffusion the aim of governmental policies is quite the opposite — 100 percent of the
population infected; hence i2010 and similar projects. In studying diseases differential equations
have been successfully used [4], but such an
approach cannot be applied here because of the
lack of necessary data.
Data used in this article are the yearly statistics collected by Eurostat, the Statistical Office
of the European Communities [5]. For each data
set a logistic growth function is constructed in
the form
y (t ) =
a
(
)
.
F
100%
1
2
3
50%
4
(1)
Here y is the penetration rate of broadband
Internet access (or percentage of households/
enterprises with such access) at time t. Constants
a, b, c, and d are time-independent growth
parameters: a is the growth limit, and b, c, and d
are the time-scaling factors. These represent the
collective effect of all factors (economic, technological, social, etc.) that influence the diffusion
process. Their numerical values are determined
using the method of least squares fitting to the
measured data in hand. This forces the plot of
the growth function y(t) to pass as close to every
data point as possible. As the function contains
four unknown growth parameters, the data set
should contain not less than four data points.
Otherwise, the values of some parameters have
to be assumed.
The popularity of broadband Internet increases from year to year. Each time a new data point
is taken into account in the model, the growth
parameters are to be determined anew, and the
future growth path including the growth limit
may change slightly. The limit becomes a moving
limit, per analogy to moving average known from
statistics.
DISCUSSION
There are no ideal models; potential sources of
modeling errors are [6]:
• The model fails to include significant variables.
• The model includes superfluous variables.
• The model uses wrong data.
• The model assumes the wrong function.
The first two items do not apply here because
the model is descriptive and does not touch the
cause-effect relations.
The data concern the broadband Internet,
but broadband represents various numerical values in various countries [5], and this may influence comparisons among countries. With respect
to the last item from the list above, Eq. 1 is
quite appropriate, and the figures below show
0%
2000
2005
2010
2015
Year
Figure 1. Diffusion of broadband Internet in Poland for various user groups. 1:
percentage of large enterprises (>250 employees) with broadband Internet; 2:
the same for enterprises with 50 to 249 employees; 3: the same for enterprises
with 10 to 49 employees; 4: percentage of households with broadband Internet;
5: penetration rate (the number of broadband access lines per 100 inhabitants). The points represent the observation data [5], and lines the results of
author's calculations following Eq. 1. The limits (asymptotes) of lines 3, 4,
and 5 are noticeable after 2012.
that it fits the data in hand well. One should
mention, however, that some researchers use
more sophisticated models; the interested reader
is referred to R&D report [7].
In determining the model parameters, one
can impose additional constraints on the logistic
function. Indeed, in a few cases, when the calculated trend line tended to exceed 100 percent,
we added an arbitrary requirement that it must
not surpass that percentage.
EXAMPLES:
LIMITS, DELAYS, AND GAPS
Due to limited space, it is impractical to examine every EU country here; therefore, only
three countries are presented as examples; similar patterns are observed in other countries.
These countries are Poland, Romania, and
Estonia. Poland is the sixth largest country in
the EU in terms of population size, GDP, and
area. Its population is 38.1 million, the GDP
per capita is 58 percent of the EU average, and
the Human Development Index (HDI) equals
0.880. The population of Romania is 21.5 million, its GDP per capita reaches 46 percent of
the EU average, and the HDI is 0.837. The
population of Estonia is 1.3 Mmillion, its GDP
per capita amounts to 68 percent of the EU
IEEE Communications Magazine • April 2010
IEEE
BEMaGS
5
1 + exp •–b + c d + t —˜
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
53
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
100%
1
2
3
4
50%
5
6
0%
2000
2005
2010
2015
Figure 2. Broadband Internet diffusion in Estonia (continuous lines) and
Romania (dashed lines). 1 and 2: percentage of large enterprises connected to
a broadband network; 3 and 4: percentage of households connected to a
broadband network; 5 and 6: the penetration rate of broadband. The points
represent the observation data [5], and lines the results of author's calculations following Eq. 1.
average, and the HDI is 0.883. As concerns the
broadband Internet penetration rate, Poland
ranks 26th, Romania 23rd, and Estonia 10th
among the EU countries.
LIMITS
Figure 1 illustrates the diffusion process of
broadband Internet in Poland. The figure compares the growth of Internet diffusion in enterprises and households. At first glance data series
3, 4, and 5 seem to grow without limits, and only
application of Eq. 1 makes it possible to infer
their limits. The percentages of households and
small enterprises with broadband Internet exhibit the lowest limits; the same relations are
observed in all countries.
DELAYS
Figure 1 shows that the growth lines are delayed
relative to one other. By definition, the delay
6t(y) is the time difference between the two lines
at a given growth level y. It can be determined
graphically or using function t(y), the inverse
function for y(t) from Eq. 1. For example, the
delay of line 3 relative to line 2 at the level of 50
percent is about four years. Note, however, that
the delay cannot be determined for levels above
60 percent. The reason is that line 3 does not
reach such high values: its growth limit is lower.
GAPS
Figure 1 illustrates the degree of digital divide
among various user groups. For large enterprises
(> 250 employees) the percentage limit is at 94
percent. For medium ones (50 to 249 employ-
54
Communications
IEEE
A
BEMaGS
F
ees) it is at 77 percent, and for small enterprises
(10 to 49 employees) at 53 percent. For households, the penetration limit is the lowest and
equals 51 percent.
Once the growth limits are known, it is
straightforward to determine whether the digital
gap between two given user groups or countries
will vanish, or not. The gap, 6y(t), is defined as
the difference between two growth functions, y1
and y2, at a given reference time t. For instance,
in Fig. 1 the gap between lines 3 and 2 in 2005
reaches 30 percentage points or so. In this example the lines do not converge; the process that
started at a lower level remains always delayed,
and the gap remains forever: it is irreducible.
However, in the case of lines 3 and 4, the gap
first increases (up to 2006 or so) and then
decreases asymptotically.
Generally, the gap varies until it reaches its
asymptotic value. It vanishes only if the two
growth lines have a common limit. As the growth
lines approach the limit asymptotically, the process of closing the gap takes a long time, theoretically infinite.
The irreducible gap is the limit to which the
gap 6y approaches when time tends to infinity. It
is easy to notice that it equals the difference |a1
– a2|, where indices 1 and 2 differentiate between
the growth lines, and a has been introduced in
Eq. 1. For instance, in the figure the irreducible
gap between 1 and 3 is about 40 percentage
points.
FURTHER EXAMPLES
Figure 2 compares the growth of broadband
Internet popularity in two other EU countries,
Estonia and Romania. (To make the figure readable, only two user groups are shown: households and large enterprises.)
For large enterprises in Estonia, the limit
exceeds 95 percent, whereas in Romania it is
lower, about 75 percent. For households in
Estonia the limit is about 65 percent; in Romania it is also lower, about 25 percent. In spite of
these differences, the penetration rates (that
merge households and enterprises) in both
countries tend to the same limit of about 25
percent.
DISCUSSION
In the long term, each and every household will
have broadband access to the Internet, as we all
wish. Consequently, it appears that the ultimate
limit for the percentage of households with
Internet access should be 100 percent. However,
it is difficult to reach such a high limit, as there
are people immune to the Internet, afraid of
cybercrime, or those below the poverty line (we
will come back to these issues later on).
The authors of [7], for instance, account for
this fact by assuming the limit (they call it saturation level) of broadband Internet penetration
rate at 20 percent for Poland and 55 percent for
France. At the time of their study, broadband
services in the EU were at an early phase, and
the diffusion data were scarce, so they had to
assume some values. In contrast, in this article
the limiting values below 100 percent result
from the actual data on broadband Internet diffusion [5].
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
In the future, with the mass applications of
the broadband Internet of Things, the limit may
surpass 100 percent, as has already happened
with the telephone penetration rate in some
countries.
0%
The theoretical limits, calculated for each of the
27 EU countries, are presented in Fig. 3, together with the actual penetration rates for each EU
country. These limits are expected to be reached
in the long-term perspective if the present development conditions continue in each country without change. The limits are the highest in Malta,
Sweden, Germany, and Denmark, but even there
they are well below 90 percent, the i2010 target;
the actual penetration rates are even lower.
This indicates that the i2010 target of 90 percent penetration rate, uniform across the whole
EU, may not be reached easily. Reaching it
requires significant efforts; we will come back to
that issue later on.
Denmark
TASKS
The fact that half of the people surveyed do not
need or want broadband access at home may
indicate that a significant part of society has
30%
40%
50%
Finland
Hungary
Netherlands
Norway
Belgium
United Kingdom
Estonia
Luxembourg
France
Spain
Cyprus
Romania
Ireland
Austria
Slovenia
Bulgaria
Latvia
Italy
Portugal
Lithuania
Czech Republic
Poland
Slovakia
Penetration rate 2007
Limit to growth
Greece
0%
10%
20%
30%
40%
50%
Figure 3. The current diffusion limits in the EU countries (author's calculations
following Eq. 1) compared with the actual (2007) penetration rate of broadband Internet. Source of the penetration data: [5].
another hierarchy of needs and values, and does
not know, does not understand, and/or does not
appreciate the benefits such access can offer.
Bottlenecks are households and small enterprises in rural areas and poor social strata. To
change this attitude, additional stimulus pro-
IEEE Communications Magazine • April 2010
Communications
IEEE
F
Sweden
Germany
BARRIERS
20%
BEMaGS
Malta
THE I2010 PROJECT
It follows from Fig. 3 that a significant part of
the population in the European Union does not
fully participate in the Information Society. This
will continue as long as the present Internet
development conditions do not change. Migration to broadband Internet in urban areas and
large enterprises is developing well. It is, however, delayed in small enterprises, households, and
especially in rural areas where a large part of a
country’s population lives.
The European Commission comments on this
as follows:
“Despite the general increase in broadband connectivity, access in more remote and rural regions is
limited because of high costs due to low density of
population and remoteness. Population scarcity
limits the exploitation of economies of scale, entails
lower rates of demand and reduced expected returns
from investment. Remoteness often implies the need
of bridging longer distances from the local
exchanges to the premises and to the backbone.
Commercial incentives to invest in broadband
deployment in these areas often turn out to be
insufficient. On the positive side, technological
innovation is reducing deployment costs.” [8]
In line with these comments, the EU Commission recently proposed an extra €1 billion in
aid to stimulate the spread of broadband Internet [9].
There are numerous reasons for such disproportions, and they may be different in each
country. These may be cultural, language, disability, age, and gender barriers, lack of skills,
precarious economic conditions, and so on. An
EU survey has identified the major reasons cited
for not having broadband Internet at home. The
most significant are: “not needed or not wanted,” “too high costs,” “lack of skills,” and “privacy or security concerns” [5]. The first one was
indicated by about 50 percent of responders.
10%
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
55
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
grams and resources (effort, capital, etc.) are
required. Favorable conditions need to be created in various areas [3]:
• Content area (to address the real needs of
small enterprises and rural communities)
• Education/promotion area (to educate citizens at all levels, stimulate involvement of
local communities)
• Investment area (public/governmental infrastructure sharing)
• Legal/regulatory area (remove legal barriers; encourage new business models; competition, investments, and sharing
infrastructures; facilitate fiscal conditions;
broadband access as universal service; solve
issues of spectrum access and intellectual
property rights)
• Business area (public aid; new business
models)
0%
5%
10%
15%
20%
Germany
Italy
France
United Kingdom
Spain
Netherlands
Greece
Portugal
Czech Republic
Hungary
Belgium
Bulgaria
Austria
Sweden
Slovakia
Finland
Denmark
Ireland
Lithuania
Latvia
Slovenia
Estonia
Cyprus
Malta
5%
10%
15%
20%
Figure 4. Market index — increase of broadband subscriber base required in
each EU country to reach the i2010 target; European Union = 100 percent.
(Author's calculations.)
IEEE
• Technology area (cheap/free software and
inexpensive hardware)
• Organization area
QUANTIFYING EFFORTS
The creation of the single European information space means significant efforts must be
made in each European country, and each
country’s position is specific. At first glance it
seems that countries in which the current penetration rate is the lowest confront the greatest
effort to reach the i2010 target. However, that
effort is to be measured not in terms of the
penetration rate but in terms of the increase of
the broadband subscriber base required to
reach the target.
To quantify that effort, we propose the market index, defined as the product of the country’s population and the required increase in
the actual penetration rate to reach the 90 percent target. It is the potential market size
induced by the i2010 project (the size of the
new subscribed base to be created). In Fig. 4 it
is normalized in such a way that 100 percent
represents the whole European Union. Seven
countries (Germany, Italy, France, United
Kingdom, Spain, Poland, and Romania), represent a major part of the new subscribers part of
the i2010 target.
CONCLUSIONS
Luxembourg
Communications
F
The EU countries differ not only in the market
size mentioned above, but also in wealth, or the
GDP (PPP) per capita, and these two factors
play major roles in ensuring uniform broadband
Internet penetration. Efforts needed to reach
the i2010 target increase with the size of the
base of new broadband subscribers and decreases with the country’s GDP: the greater the subscribers’ base to be created and the lower the
GDP, the greater the efforts. To make reasonable comparisons among the EU countries in
this aspect, an effort index is proposed. It is
defined as the ratio of the market index to the
GDP (PPP) per capita. It is shown in Fig. 5 for
each EU country.
Bottlenecks are to be expected to materialize in poor social strata, rural, areas and small
enterprises in countries facing the greatest
efforts. These need special attention to avoid
significant delays in reaching the i2010 target.
Figure 5 indicates that seven countries (Poland,
Germany, Romania, Italy, Spain, United Kingdom, and France) are facing the greatest
efforts; consequently, they require special assistance. A question arises on how to allocate that
aid to lower the probability of major bottlenecks before they materialize. A rational guide
would be to distribute it proportionally to the
effort index.
Romania
56
BEMaGS
QUANTIFYING BOTTLENECKS
Poland
0%
A
The model proposed here matches the historical
data well. It infers intrinsic limits to growth of
the broadband Internet penetration rate that
may be unnoticeable otherwise. These growth
limits depend on many factors. Rural and poor
regions and small enterprises are major bottlenecks.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
The model does not confirm that the existing
disproportions in broadband Internet penetration rates between countries, regions, and user
groups vanish with time. They increase and
decrease with time and, in some cases, are irreducible and will exist as long as the present
development conditions continue. This warning
calls for a review of our attitude toward reducing
the digital divide not only in the European
Union but also elsewhere.
The i2010 target of 90 percent penetration
rate, uniform across the whole EU, may not be
reached as quickly as originally expected. This
target requires significant efforts, and some
countries/regions require special assistance, such
as the recent program [9]. The effort index could
serve as a guide for where that assistance should
be directed.
The approach used here to examine Internet
diffusion in the European Union may also be
applied to non-EU countries as well as to other
services besides broadband.
ACKNOWLEDGMENTS
The author would like to thank Juerg Daellenbach, Sean Moore, Marcin Struzak, and anonymous reviewers for their contributions and
comments, which have strengthened this text.
0%
5%
10%
15%
5%
10%
15%
A
BEMaGS
F
Poland
Germany
Romania
Italy
Spain
United Kingdom
France
Bulgaria
Hungary
Portugal
Czech Republic
Greece
Netherlands
Slovakia
Belgium
Austria
Sweden
REFERENCES
[1] Commission of the European Communities, “Preparing
Europe’s Digital Future: i2010 Mid-Term Review,”
COM(2008) 199, Apr. 2008.
[2] “Opinion of the Committee of the Regions on Bridging
the Broadband Gap and i2010 eGovernment Action
Plan (2007/C 146/09),” Official J. EU, June 30, 2007.
[3] R. Struzak, “Growth of Broadband Internet in Poland —
Models, Trends, and Limits,” Telekomunikacja i Techniki
Informacyjne, vol. 2009, no. 1–2, pp. 38–48, (in Polish);
also: R. Struzak, “Broadband Internet Access: Trends
and Limits,” Proc. 4th BroadBandCom ‘09, Wroclaw,
Poland, July 15–18, 2009.
[4] M. Keeling, “The Mathematics of Diseases,” Plus Mag.,
Mar. 2001, p. 3–8.
[5] EUROSTAT databases, accessed Mar. 22, 2009; ____
http://
epp.eurostat.ec.europa.eu
______________
[6] R. I. Ackoff, Scientific Method: Optimizing Applied
Research Decisions, Wiley, 1962.
[7] R. Montagne et al., “Broadband Access Roadmap Based
on Market Assessments and Technical-Economic Analysis,” BROADWAN, Deliv. D15, 2005.
[8] Commission of the European Communities, “Communication from the Commission to the Council, the European Parliament, the European Economic and Social
Committee, and the Committee of the Regions: Bridging The Broadband Gap,” COM(2006) 129, Brussels,
Belgium, Mar. 20, 2006.
[9] European Commission, “Better High-speed Internet
Access Needed to Revitalise Europe’s Rural Regions,”
press rel. no. IP/09/343, EU Brussels, Mar. 3, 2009.
BIOGRAPHY
RYSZARD STRUZAK [LF] (r.struzak@ieee.org)
__________ is a full professor
at the National Institute of Telecommunications, Poland,
and Wroclaw University of Technology, and co-director of
the ICTP School Series on Wireless Networking. He is the
author/co-author of some 200 publications and 10
Denmark
Lithuania
Latvia
Finland
Ireland
Slovenia
Estonia
Cyprus
Malta
Luxembourg
0%
Figure 5. Effort index of each EU country to reach the i2010 target. European
Union = 100 percent. (Author's calculations.)
patents. He was the former acting assistant director and
head of the Technical Department CCIR-ITU, Editor-in-Chief
and Editorial Board Chair of Global Communications, and
a consultant to ITU, UN-OCHA, WB, IUCAF, and other entities. He is co-founder and former Chair of the International Wroclaw Symposium on EMC. He was elected to leading
positions in ITU-RRB, CCIR, URSI, and CISPR. He is the
recipient of the ITU Silver Medal, two International Symposia awards, and national awards and decorations. He is
a member of the New York Academy of Science and an
Academician of the International Telecommunication
Academy.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
57
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
Service Traffic Management System for
Multiservice IP Networks:
Lessons Learned and Applications
JungYul Choi, Seung-Hoon Kwak, Mi-Jeong Lim, Taeil Chae, Byoung-Kwon Shim, and Jae-Hyoung Yoo,
KT Corporation
ABSTRACT
Next-generation networks offer new opportunities and challenges to Internet service providers as well as providers of other online
services. Service providers can now deploy new
services over an IP network infrastructure without building their own networks. In an open network environment, the network resources of
ISPs should be fairly open to third parties that
plan to launch their own services over the network. To actively respond to the changing network paradigm, it is essential to measure the
traffic of individual services, and to estimate
their cost for cost accounting between service
provider and ISP. However, current traffic measurement techniques only provide the total traffic volume in links, without reporting the
operator whose services flow through the links.
Some commercial products can classify traffic
into each application at a specific spot, but we
should install monitoring systems at every spot
throughout the entire network in order to
observe which service traffic flows in every link.
To satisfy the requirements of the NGN environment, we developed the Service Traffic Management System that can analyze the traffic of
individual services based on user log data. STMS
can report not only the traffic of individual services in every link, but also user behavior for
each service. In addition, this article shares our
experience of STMS development. We also
introduce how we utilize STMS in IP network
design, and discuss business and management
support.
INTRODUCTION
Each year Internet service providers (ISPs),
which own large-scale network infrastructure,
pump a great deal of money into building their
networks in order to sustain soaring Internet
traffic. The explosive increase in Internet traffic,
which is mostly due to peer-to-peer services and
free-rider services, places a huge burden on
ISPs, with little resulting revenue from their
network investment. On the other hand, the
emergence of new services such as Internet Pro-
58
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
tocol television (IPTV) and fixed mobile convergence (FMC) is expected to offer new added
value to ISPs and opportunities to service providers. This is because the next-generation network (NGN) environment now enables the
provision of a variety of new services over IPbased transport networks by utilizing network
and service control platforms without building a
separate network for each service [1, 2]. This is
the fruit of the realization of the NGN environment, which can pave a new way to profitable
network operation.
In this context, an exact cost accounting of
individual services is essential for ISPs that provide services over their own network, as well as
for service providers that utilize open networks
for service provisioning. Cost accounting of individual services in network building and operation will be the basis for the settlement of
accounts between ISPs and service providers, or
between business departments and a network
operating department in an ISP. Accordingly, we
should measure and analyze the traffic volume
of individual service flows in every link over the
entire network. User behavior in the use of services and traffic characteristics of services should
also be examined and utilized in network design
and planning.
However, legacy traffic measurement techniques simply provide the total traffic volume in
links, without reporting which service traffic
flows through the links [3, 4]. It does not report
on user behavior in the use of services, nor does
it report the traffic characteristics of services.
Recent commercial products can classify traffic
into each application and provide detailed traffic
characteristics [5, 6]. Deep packet inspection
(DPI)-based traffic analysis can provide application-level contents of traffic by inspecting the
payload of each packet [7]. However, application
of DPI-based techniques should be carefully
applied in large ISPs for protecting personal privacy and network neutrality [8]. These solutions
give benefits in providing detailed applicationlevel traffic classification, but have limited scalability and high cost. We should install such traffic
monitoring systems at every spot throughout the
entire network in order to exactly measure the
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
VoD servers
F
To address this
limitation of legacy
traffic measurement
techniques, we
IPTV
IP core network
Access
network
VoIP
Access
edge router
Head-end
center
developed a new
traffic management
system that can
Core
edge router
classify individual
service traffic in links
over the entire
Internet
access
network, and
IPTV VOD
analyze user behav-
IPTV multicast
ior in the use of
VoIP
services with high
Internet
economic benefit.
Figure 1. Multiservice IP network environment.
traffic volume of individual services and where
they flow [9]. Thus, to address the changed open
network environment for NGN, there is a
demand to develop a service traffic management
system for achieving the following goals:
• Can it measure the traffic volume of individual service flows in every link over the
entire network and where they flow?
• Can it provide an economic solution that
does not require traffic monitoring systems
at each spot?
• Can it provide user behavior data in the use
of services in links and regions?
• Can it provide basic data for estimating cost
accounting of each service provisioned in
an open network environment?
• Can it provide basic data for an IP network
design and planning reflecting the characteristics of individual service traffic in specific regions?
These goals accord with the requirements of
Korea Telecom (KT) to do internal trading
between business departments and the network
operating department in our Company-in-Company system. The open network environment has
similar issues in cost accounting to those when
dealing with two separate companies. To respond
to the changing network environment, and utilize network design and business support, we
developed the Service Traffic Management System (STMS). A unique feature of STMS that
differentiates it from legacy traffic management
systems relying on passive measurement is the
use of user log data of individual services when
STMS analyzes the traffic volume of individual
services in links. In this article we introduce the
detailed functions of STMS and discuss how to
compute service traffic. Verification of accuracy
of the results from STMS is also presented by
comparing them with measured traffic. We share
the experiences we gained during system development. Finally, we present a new design
methodology for IP networks and business support as possible applications of STMS.
STMS
As shown in Fig. 1, diverse service traffic flows
over IP networks, and a single link can hold
multiple different types of service traffic. Legacy
traffic measurement systems can only collect
data on incoming/outgoing traffic at network
equipment using Simple Network Management
Protocol (SNMP) [3], and analyze flow-level
traffic [4, 10] or application-level traffic [7].
While these systems are useful in analyzing
packet, flow, or application-level traffic, they can
place a big burden on ISPs if there is the need
to establish them in every link for precise service
traffic management throughout the entire network. To address this limitation of legacy traffic
measurement techniques, we developed a new
traffic management system that can classify individual service traffic in links over the entire network, and analyze user behavior in the use of
services with high economic benefit. STMS can
analyze services that require authentication processes and generate user log data such as IPTV
video on demand (VOD), IPTV real-time channel type, and Session Initiation Protocol (SIP)based VoIP.
STRUCTURE AND FUNCTIONS OF STMS
The structure and function blocks of STMS are
illustrated in Fig. 2. STMS has a three-tier architecture. A collecting server linked with network
management systems (NMS) gathers source
data. An analyzing server, which is the heart of
STMS, processes the linked source data and
computes service traffic based on the source
data. A web server provides users with a screen
on which they can inquire regarding the results
of service traffic and statistics.
STMS periodically collects network facility
data, user log data of services, and traffic data
from related network management systems using
FTP. Especially, network facility data provides
connection link information between two pieces
of equipment (port-level) for reconstituting the
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
59
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Traffic data
User behavior data
Statistics
Service traffic analysis
User behavior analysis
Failed/subscribers statistics
A
BEMaGS
F
Web server
Analyzing
server
Service traffic
VOD user behavior
Network facility
End-to-end traffic
VOD contents usage
Subscribers
Traffic trend
VoIP user behavior
Peak time
Daily user behavior
Overloaded links
Service account log
Traffic data
Collecting
server
Access
network
traffic
Core
network
traffic
VOD log
Channel
log
VoIP log
Facilities and services contract
Network
facility
Service
contract
information
NMS
IP networks
Figure 2. Configuration and functional entities of STMS.
network topology. The equipment ranges from
subscriber aggregation switches to access and
backbone network routers, and application
servers. User log data for IPTV VOD service,
for example, includes subscriber ID, set-top-box
(STB) IP address, event time, VOD contents
title, VOD server IP address, and bandwidth
(constant bit rate [CBR]-encoded, bits per second). Traffic data measured by SNMP in access
and backbone networks provides total port-level
traffic information. Table 1 summarizes the
linked source data collected by NMS.
Based on the source data, STMS analyzes the
traffic of individual services, user behavior in the
use of each service, and the statistics of the network facility and subscribers. Regarding traffic
data, STMS provides service traffic at the port,
system, and node (e.g., central office [CO]) levels. Future traffic trends based on time-series
analysis as well as traffic ratio of services in links
are provided. User behavior in the use of service
includes the average service time and the arrival
rate of each service, the ratio of voice and video
communications, and VOD contents usage rate
by areas. Statistics of the network facilities and
subscribers are also analyzed and managed at
the port, system, and node levels.
COMPUTATION OF SERVICE TRAFFIC
STMS computes the traffic of individual services with collected source data through the following procedure, illustrated in Fig. 3. After
completely collecting all source data, STMS
first reconstitutes the network topology from
access edge routers to backbone networks based
on connection link data between network equip-
60
Communications
IEEE
ment. The connection link consists of upper
equipment and its port corresponding to connected equipment, and lower equipment and its
port corresponding to connected equipment. In
addition, STMS maps subscribers (e.g., IPTV
STB, VoIP phone, and PC) onto an access edge
router because STMS regards the end of the
subscriber’s part as an access edge router of an
IP network. Next, according to the transmission
policy of each service, STMS computes end-toend service traffic individually. For doing this,
STMS configures an end-to-end session of each
service from access edge routers to application
servers using the user log data of services.
Finally, based on the subscriber’s ID for services and the Internet access line linked with
access edge routers, the computed service traffic in an end-to-end manner is mapped onto
links and nodes on the route of the session.
Computing service traffic is explained individually as follows.
For IPTV VOD service provided through
unicast from VOD servers to a subscriber’s
STB, STMS configures the VOD service session
with {access edge router’s IP address, session
start time, stop time, VOD server’s IP address,
bandwidth} from the collected user log data.
Since the original log data only provides a single
event time, we need to configure a session from
start time to stop time based on the group of
the same subscriber’s IP, STB IP, contents title,
and VOD server IP. The start event includes
start, replay, fast forward, and fast backward
(rewind), and the end event includes end and
pause. When end-to-end traffic computation of
IPTV VOD service has been finished, each ses-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Data type
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Management systems
Facility (backbone)
{equipment code (upper), port number code (upper), equipment code
(lower), port number code (lower)}
Facility (access)
{subscriber ID, equipment code}
IPTV (VOD)
{subscriber ID, STB IP, event time, contents title, VOD server IP, bandwidth},
where event = {start (replay), end, pause, fast forward, fast backward}
IPTV (channel type)
{subscriber ID, STB IP, event time, channel number, channel server IP, bandwidth}, where event = {start, end}
VoIP
{calling party number, called party number, event time, service type (voice,
video), bandwidth} , where event = {start, end}
Traffic
{equipment code, port number code, date, incoming traffic (b/s), outgoing
traffic (b/s)}
IPTV (channel type)
{equipment code, port number code, date, total channel traffic (b/s)}
Facility data
User log data
Linked source data
Traffic data
A
BEMaGS
F
Interface
FTP
Table 1. Source data collected from management systems.
sion’s traffic is allocated in a timeline to the
route of the session composing links and nodes
on the network topology. Using the time for
which the session is maintained and the bandwidth of the session, STMS computes the average traffic bandwidth for the session. Let us
exemplify how to compute the traffic volume of
this service in a specific link. It is assumed that
there are three sessions with 2 Mb/s from 0 to
30 min (session 1), 6 Mb/s from 10 to 50 min
(session 2), and 4 Mb/s from 20 to 40 min. As a
result, the one-hour average bandwidth of the
service in the link is computed as 6.33 Mb/s (=
{2 Mb/s * 30 * 60 s + 6 Mb/s * 40 * 60 s + 4
Mb/s * 20 * 60 s}/3600 s).
For IPTV real-time channel service provided
through multicast from a head-end center to an
STB, STMS configures the channel session with
{access edge router’s IP address, start time, stop
time, channel ID, channel server’s IP address,
and bandwidth}. When traffic computation of
end-to-end IPTV channel service is finished,
each channel’s traffic is allocated in the timeline
to the route of the channel composing links and
nodes on the network topology. From Fig. 4, we
can easily understand how STMS computes
IPTV channel service traffic in a link. Figure 4a
shows that seven subscribers watched three different channels, the traffic of which flowed
through the link at different times. Due to the
multicast nature of IPTV channel service, there
are only three channels’ traffic flowing through
the link, as shown in Fig. 4b. As a result, the
one-hour average bandwidth of channel service
in the link is computed as 6.33 Mb/s (= {2 Mb/s
* 50 * 60 s + 4 Mb/s * 50 * 60 s + 2 Mb/s * 30 *
60 s}/3600 s).
A VoIP service session consists of {edge
router for calling party, edge router for called
party, start time, end time, conversation type
[voice or video communications], and bandwidth}. When end-to-end session traffic is calculated based on conversation time and bandwidth,
the session traffic is allocated to the route of a
VoIP session on the topology. Very similar to
the traffic computation process for IPTV service
except for the bidirectional feature of VoIP service, STMS computes VoIP service traffic in all
links over the entire network.
Regarding Internet access service traffic,
STMS gathers the total amount of traffic in all
links. Because the Internet access service does
not have any user log data, STMS cannot configure an end-to-end session for it. Thus, after subtracting the traffic of all services from the total
traffic in links, what remains is the Internet traffic.
VERIFICATION AND RESULTS OF
SERVICE TRAFFIC COMPUTED
The most critical point in the success of STMS
development is how accurately the service traffic
computed from user log data reflects the real
traffic. The possible sources of error in STMS
can be classified to one of the following two categories:
• Errors and omissions in source data, such as
traffic data, user log data, and network
facility data
• Errors in algorithms to compute traffic of
individual services
From checking the possible errors, it was
found that 1 percent of log data errors are due
to missing user log data and mismatch of subscriber’s contract data, and about 5 percent of
network facility data errors are due mostly to
missing data. Errors in service traffic computing
algorithms can be verified from the following
comparison between STMS traffic and measured
traffic. Figure 5 compares the IPTV VOD service traffic in a day from STMS with measured
traffic at VOD server farms located in each
node (CO). Each node shows 24 hours of traffic
data. In most nodes the computed traffic from
STMS is the same as the measured traffic within
a 5 percent error range. The difference between
them is revealed mostly due to errors and loss in
network facility data. In addition, to verify the
algorithm to compute IPTV real-time channel
traffic, we tapped traffic at a 1 Gb/s link using a
passive measurement system, and classified the
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
61
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
An interesting result
regarding user
behavior in the use
Head-end center
(VOD server)
IPTV STB
Facility data
VoIP
(called party)
VoIP
(calling party)
of VoIP service is
that average holding
IDC/PC
(web server)
Internet access
time (AHT) between
(a)
VoIPs is much longer
than that between
VoIP and PSTN, and
User log data
between VoIP and
mobile phone. AHT
(b)
from VoIP to mobile
phone is reported
the shortest time.
Traffic data
(c)
IPTV traffic
VoIP traffic
Internet traffic
Figure 3. Service traffic computing process: a) reconstitution of network topology from network facility data;
b) end-to-end service traffic computation from user log data; c) allocation of end-to-end service traffic over
the entire network.
channel traffic for comparing STMS traffic. The
VoIP traffic computing algorithm was verified by
comparing STMS traffic with the measured traffic of VoIP-level quality of service (QoS) at a
differentiated services (DiffServ)-enabled router.
From this, we can reason that the service traffic
computed by STMS based on user log data will
be the same if the integrity of network facility
data and user log data is guaranteed.
We add some results of service traffic
obtained from STMS. Figure 6 shows the total
traffic volume and each service traffic volume
summing up at access edge routers in a day. At
the peak time of the network, the Internet access
service occupies about 57 percent of total traffic
volume, IPTV VOD 9 percent, IPTV channel 34
percent, and VoIP 0.1 percent. The majority of
total traffic volume is still due to the Internet
access service, but IPTV service traffic has quickly soared in the KT network since KT deployed
IPTV services in 2007. When it comes close to
the backbone network, the effect of IPTV channel traffic gradually decreases due to the feature
of multicasting transmission on which IPTV
channel service relies. VoIP traffic currently
occupies a very small portion of total traffic volume, mostly due to the narrow bandwidth of
each session, but is fast growing thanks to
increased subscribers.
STMS also reports some interesting results
regarding the traffic volume per user of services.
At the peak time of individual services, the traffic volume per user of IPTV VOD service is two
times that of Internet access service. IPTV channel service produces five times larger traffic vol-
62
Communications
IEEE
ume than Internet access service. In other words,
in terms of traffic volume, the cost of IPTV
channel service is five times that of Internet
access service. This result gives the basis for the
settlement of accounts between business departments and network operating departments with
the totally occupied traffic volume of each service end-to-end throughout the entire network.
Traffic volume per user of VoIP service is negligible compared to other services. An interesting
result regarding user behavior in the use of
VoIP service is that average holding time (AHT)
between VoIP users is much longer than that
between VoIP and the public switched telephone
network (PSTN), and between VoIP and mobile
phones. AHT from VoIP to mobile phone is
reported to be the shortest. User behavior
regarding AHT seems mostly due to the different chargea for the types of calls.
LESSONS LEARNED FROM THE
DEVELOPMENT OF STMS
In the beginning of the STMS development project, the prime obstacle we faced was that we
were not sure of the accuracy of service traffic
computed by STMS compared to real traffic
data. To the best of our knowledge, there is no
attempt to compute service traffic as STMS does
in academia or industry. As introduced earlier,
we had to focus on the verification of service
traffic computed. From our experience during
the verifying process, we found valuable lessons
for improving the quality of STMS.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
During verification of traffic computation
of IPTV VOD service, we found that there
were missing items in user log data that are
indispensable in computing service traffic.
Since KT’s IPTV VOD service is provided by
streaming, a user can select start, pause,
replay, and stop as well as fast forward (FF)
and fast backward (FB). Initial log data only
contains the first four events, but the latter
two events are not provided because these
events are not indispensable for service provisioning. Thus, there was missing traffic
between the time of FF/FB to stop (or pause).
Thus, we asked the IPTV VOD service department to add the two events in the log data,
and STMS consequently produces more accurate results. Similarly, there was only the start
time of the channel but not its end time in
user log data of IPTV channel service. The
end time of the channel can be regarded as
the start time of a new channel by the same
user’s choice at the same time. However, the
problem occurs when a user turns off an STB
without reporting the end time of the channel.
Currently, the IPTV server periodically performs a health check of the STB, so we may
guess the end time of the channel will be
halfway between the turning off of the STB
and the health check. Such an inaccurate log
data possibly produces errors in producing service traffic. To resolve the problem, we asked
to add the end time of a channel in user log
data and applied this in STMS. Likewise, the
integrity of user log data is the most precious
thing for success of STMS. Thanks to the verifying process for STMS functions, we found
missing log data and errors in the STB for the
log data generating function.
There is another issue of integrity for network facility data. Errors and missing data for
the network facility are possible in the area
between access and backbone networks due to
different ownership of network management.
Since KT owns and manages hundreds of thousands of switches and routers for IP networks,
KT has to divide the network into access and
backbone with different network management
systems. Thus, we had to check the integrity of
the network facility data located in the shared
area managed at different NMS. We then reported the erroneous data to the NMS for fixing and
updating them by authorizing the managing task
for related equipment.
Finally, we have a comment about management lag time in STMS. Due to the log data
generation by NMS and difficulty of real-time
data collecting, STMS collects source data from
NMS daily. In addition, many sessions start just
before 0 o’clock and end after 0 o’clock the following day; thus, STMS can generate the sessions when it collects the log data after another
day. It also takes around 7–8 hours for STMS to
generate end-to-end session traffic of services
and compute service traffic in all links over the
entire network based on the source data reaching around 10 Gbytes. As a result, STMS can
only report the traffic data around two days after
the real date. In order to minimize the management lag time, it is recommended to collect the
source data from NMS every hour and have
Subscriber 7
0
IEEE
F
Channel 3 (4 Mb/s)
Channel 2 (2 Mb/s)
Channel 2 (2 Mb/s)
Channel 1 (2 Mb/s)
Channel 1 (2 Mb/s)
Channel 1 (2 Mb/s)
10
20
30
40
50
60 (min)
(a)
Channel 3 (4 Mb/s)
Channel 2 (2 Mb/s)
Channel 1 (2 Mb/s)
0
10
20
30
40
50
60 (min)
(b)
8 Mb/s
6 Mb/s
4 Mb/s
2 Mb/s
6.33 Mb/s
0
10
20
30
40
50
60 (min)
(c)
Figure 4. Example of how to compute the average traffic bandwidth of IPTV
channel service in a link: a) channel traffic from user log data; b) multicast
channel traffic; c) average traffic bandwidth.
STMS process them on an hourly basis. However, it is practically very difficult to collect source
data from NMS every hour while keeping the
integrity of source data as well as the performance of STMS high.
APPLICATIONS OF STMS
As it provides traffic-related data, user behavior
data, and facility statistics, STMS has an important role in a data warehouse for IP network
infrastructure in KT. STMS can be expanded to
applications such as traffic monitoring, network
facility management, network design, and business and management support, including marketing promotion and cost accounting for
services. This section introduces two applications
of STMS in IP network design, and in business
and management support.
IP NETWORK DESIGN
IP network design in a multiservice environment
should reflect the traffic characteristics and user
behavior of services. However, a legacy network
design relying on total link traffic data is hard to
apply in a new environment [11, 12]. We thus
developed a new IP network design methodology
based on end-to-end traffic data provided by
STMS.
The IP network design process first gathers
IEEE Communications Magazine • April 2010
Communications
BEMaGS
Channel 3 (4 Mb/s)
Subscriber 6
Subscriber 5
Subscriber 4
Subscriber 3
Subscriber 2
Subscriber 1
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
63
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
10,000
STMS traffic
Measured traffic
8000
Mb/s
6000
4000
BEMaGS
F
Changes in the traffic route due to factors such
as routing policy change as well as changes in
network topology can easily be reflected in the
network design by allocating the end-to-end traffic of each service on the changed routes. Finally, we can accurately estimate the cost pricing of
services in network building and operation,
because we consider the individual service-related characteristic in the network design.
BUSINESS AND MANAGEMENT SUPPORT
2000
0
A
B
C
D
Nodes
E
F
Figure 5. Comparison of STMS traffic and measurement traffic.
network facility data, traffic data, user behavior
data, service demands, and routing policy including service transmission route. Next, it configures the target network topology and computes
network design parameters based on the gathered data. The following process explains how to
produce the estimated traffic volume at the target year in each link, based on end-to-end traffic
and network design parameters:
• For services that provide user log data such
as IPTV and VoIP, computing end-to-end
traffic per user of service, based on end-toend traffic that STMS provides and the
number of users at an end node (endpoint).
• Forecasting end-to-end traffic per user of
each service at the target point of time.
• Calculating end-to-end traffic of each service by multiplying forecast traffic by future
demands.
• Producing offered traffic at links by allocating the end-to-end traffic to the service
traffic routes, service by service.
• Producing designed traffic at links by considering traffic variation over time, because
the offered traffic is produced based on the
average peak time of month. To compensate for the traffic difference at the average
peak time of month and day, we compute
an adjustment factor and apply it to compute the designed traffic from the offered
traffic.
• Performing link and node design that can
accommodate the designed traffic.
On the other hand, since the Internet access
service does not have end-to-end traffic data, we
need to compute this traffic link by link, not end
to end, and add the link traffic to the service
traffic computed earlier. The capacity planning
we design is only related to the downstream traffic since the traffic volume is much higher than
upstream.
The proposed design methodology based on
end-to-end traffic provides the following features. As we consider user behavior for individual services and the traffic characteristic of each
service, it is possible to design networks more
accurately by reflecting area-specific features.
64
Communications
IEEE
A
Since KT merged with Korea Telecom Freetel
(KTF), an affiliated company of KT in mobile
wireless communications, in June 2009, KT has
been preparing to transition the company structure to a Company-in-Company (CIC) system.
Under the CIC system, business departments
and the network operating department have to
account for internal trading using network
resources for individual services. Thus, an exact
cost accounting between companies becomes a
critical issue. The open network environment
has similar issues in cost accounting between two
separate companies. Currently, cost accounting
is mostly performed based on the number of
subscribers, revenue per subscriber, and traffic
volume at the point of access between two ISPs.
There is no clear criterion in the cost estimation
of individual services from the viewpoint of network building and operation [13]. Thanks to
STMS, which provides precise information on
the traffic volume for individual services on
every link in the entire network, we can now
estimate the cost of each service based on the
traffic volume actually used by users.
Cost pricing of services is just one example of
what STMS can support. There are wide applications of STMS in business and management support. User behavior in the use of services can be
utilized as the basic data to establish a business
and management strategy, and to support a marketing plan for service promotion, new service
development, service pricing policy, and so on.
Traffic-related information can be utilized to
support decision making by management, as well
as network operating system improvement.
CONCLUSION
STMS is a useful tool for building a data warehouse for IP network infrastructure in a multiservice environment for KT. In addition to
service traffic monitoring and management, we
can expect the following merits from STMS in
various areas of applications. First, network
design can become more accurate and flexible,
because it can consider area- and service-specific
data, and easily reflect changes in traffic routes
and network topology. Second, the cost of individual services in network building and operation can be more accurately estimated. Third,
knowledge of user behavior in the use of services
in specific areas is helpful when planning business and marketing strategy. In conclusion, during this era of open networking and network
convergence, KT’s experience in developing
STMS is expected to offer a new way for network operation, and a new business strategy for
global ISPs facing the changed network
paradigm.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
REFERENCE
BIOGRAPHIES
J UNG Y UL C HOI (passjay@kt.com)
_________ received his B.S. degree
from Inha University, South Korea, in 2000, and his M.S.
and Ph.D. degrees from Information and Communications
University (currently merged with Korea Advanced Institute
of Science and Technology), South Korea, in 2002 and
2006, respectively. He has been working at the Network
R&D Laboratory, KT Corporation (formally, Korea Telecom)
since 2006. He has authored around 20 reviewed technical
journal papers, and holds around 10 patents in telecommunications networks. He was nominated for Marquis
Who’s Who in the World 2009 and International Engineer
of the Year for 2010 from IBC. He has been a reviewer of
technical conference and journal papers for IEEE INFOCOM,
IEEE Communications Letters, IEICE Transaction on Communications, IEICE Transaction on Information and Systems,
Elsevier Journal of Visual Communication and Image Representation, and ETRI Journal. His research interests are in
next-generation networks, future networks, wired/wireless
convergence, and network economics.
S EUNG H OON K WAK (shkwak@kt.com)
_________ received his B.S. and
M.S. degrees from Chonbuk National University, South
Korea, in 1995 and 2001, respectively. He has been working at the Network R&D Laboratory, KT Corporation since
1995. His research interests are in next-generation networks, network dimensioning, and traffic analyzing.
M I -J EONG L IM received her B.S. and M.S. degrees from
Chungnam National University, South Korea, in 1991 and
1994, respectively. She has been working at the Network
2200
2000
1800
1600
Total traffic
Internet access
IPTV channel
IPTV VOD
VoIP
1400
Gb/s
[1] K. Knightson, N. Morita, and T. Towle, “NGN Architecture: Generic Principles, Functional Architecture, and
Implementation,” IEEE Commun. Mag., vol. 43, no. 10,
Oct. 2005, pp. 49–56.
[2] C.-S. Lee and D. Knight, “Realization of the Next-Generation Network,” IEEE Commun. Mag., vol. 43, no. 10,
Oct. 2005, pp. 34–41.
[3] C. Fraleigh et al., “Packet-Level Traffic Measurements
from the Sprint IP Backbone,” IEEE Network, vol. 17,
no. 6, Nov. 2003, pp. 6–16.
[4] J. Quittek et al., “Requirements for IP Flow Information
Export (IPFIX),” IETF RFC 3917, Oct. 2004.
[5] Procera PacketLogic; http://www.proceranetworks.com/
products.html
________
[6] Blue Coat Packetshaper; http://www.bluecoat.com/
products/packetshaper
____________
[7] P.-C. Lin et al., “Using String Matching for Deep Packet
Inspection,” IEEE Computer, vol. 41, 2008, pp. 23–28.
[8] S. Jordan, “Some Traffic Management Practices Are
Unreasonable,” Proc. 18th ICCCN ‘09, Aug. 2009.
[9] M. Sidibe and A. Mehaoua, “Service and Network Monitoring Support for Integrated End-to-End QoS Management,” Proc. IEEE Net. Ops. Mgmt. Symp. Wksp.,
Apr. 2008, pp. 132–37.
[10] Cisco Netflow; http://www.cisco.com/
[11] T. Jensen, “Network Planning — Introductory Issues,”
Telektronik, vol. 3/4, 2003, pp. 9–46.
[12] T. Jensen, “Network Strategy Studies,” Telektronik,
vol. 3/4, 2003, pp. 68–98.
[13] G. Davies, M. Hardt, and F. Kelly, “Come the Revolution — Network Dimensioning, Service Costing and
Pricing in a Packet Switched Environment,” Telecommun. Policy, vol. 28, 2004, pp. 391–412.
1200
1000
800
600
400
200
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Hour
Figure 6. Results of service traffic computation.
R&D Laboratory, KT Corporation since 1994. Her research
interests are in traffic monitoring/management and network engineering.
T AEIL C HAE (tichae@kt.com)
_________ received his B.S. and M.S.
degrees from Yonsei University, South Korea, in 1987 and
1989, respectively, and his Ph.D. degree from Information
and Communications University, South Korea, in 2007. He
has been working for Network R&D Laboratory, KT Corporation since 1993. He has authored a few technical journal
papers in areas of optical communications networks, and
has several patents in optical systems and telecommunications. His research interests are in next-generation networks, optical packet networks, fixed mobile convergence,
and network economics.
BYOUNGKWON SHIM (bkshim@kt.com)
_________ received his B.S. and
M.S. degrees from Hanyang University, South Korea, in
1985 and 1987. He has been working at the Network R&D
Laboratory, KT Corporation (formally, Korea Telecom) since
1987. As a director and project leader, his research interests are traffic analysis and engineering for next-generation
networks, and network economics.
JAE-HYOUNG YOO [M] (styoo@kt.com)
________ is a vice president and
group leader in the Network Strategy Research Group, Network R&D Laboratory, KT Corporation. He received his B.S.,
M.S., and Ph.D. degrees from the Electronic Engineering
Department of Yonsei University, South Korea, in 1983,
1985, and 1999, respectively. Since he joined KT in 1986,
he has worked on the research and development of various
networks, QoS management and traffic engineering systems including PSTN, ATM, and fixed and mobile Internet.
His research interests include routing algorithms, traffic
engineering, fixed and mobile IP network architecture, and
next-generation operation support systems (NGOSS). He
was an Application Session Co-Chair of NOMS 2004 and a
Special Session co-chair of APNOMS 2006 and 2009. He is
an editorial board member of the International Journal of
Network Management and Journal of Telecommunications
Management.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
65
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
TOPICS IN DESIGN & IMPLEMENTATION
Safety Assurance and Rescue
Communication Systems in High-Stress
Environments: A Mining Case Study
Prasant Misra and Salil Kanhere, The University of New South Wales
Diethelm Ostry, CSIRO ICT Centre
Sanjay Jha, The University of New South Wales
ABSTRACT
Effective communication is critical to the success of response and rescue operations; however,
unreliable operation of communication systems
in high-stress environments is a significant obstacle to achieving this. The contribution of this
article is threefold. First, it outlines those common characteristics that impair communication
in high-stress environments and then evaluates
their importance, specifically in the underground
mine environment. Second, it discusses current
underground mine communication techniques
and identifies their potential problems. Third, it
explores the design of wireless sensor network
based communication and location sensing systems that could potentially address current challenges. Finally, preliminary results are presented
of an empirical study of communication using a
WSN constructed from commercially available
wireless sensor nodes in an underground mine
near Parkes, New South Wales, Australia.
INTRODUCTION
Communication systems relying on wireless links
have become integral to industry and to our
daily life. They now form a core infrastructure
component, which has led to great improvements in convenience, productivity, and safety.
Their success has led to a desire to make their
capabilities available reliably in all environments
of commercial, industrial, and social importance.
Some of these environments inherently present
challenging technical problems, which constrain
wireless communications. For example, wireless
communication between mobile devices inside
buildings and factories must often operate in
conditions of high signal attenuation, electrical
interference, and multiple reflections or echoes,
which restrict range and performance. Apart
from those requirements, which arise in specific
applications, reliable operation requires that a
communication system should be designed to
survive foreseeable accidents and emergencies.
These two situations might together be
66
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
termed high-stress environments: environments
in which by their nature it is intrinsically difficult
to provide communications, and the extreme or
abnormal conditions following an accident or
disaster, which can both destroy system components and radically alter environmental conditions in which the system must operate. We shall
focus in this article on one such environment,
which is both physically harsh under normal
operating conditions, and has significant risk of
accidents that can damage communications
infrastructure and disrupt communications:
underground mines.
Underground mines are typically extensive
labyrinths of long (perhaps several kilometers)
and narrow (only a few meters in width) tunnels.
They may employ hundreds of mining personnel
working at one time under extreme environmental conditions and distributed throughout the
mine. The overall mining process is highly mobile,
and mining machinery has to be repositioned as
the mining operation progresses; consequently,
the communication environment continually
changes. The combination of ever-changing
ground conditions with a dynamic mining system
generates a broad profile of risks, which results in
human casualties in mine accidents [1].
Management of the hazards in underground
mines requires continuous monitoring of critical
information: the presence and concentration of
flammable and toxic gases and dust, the structural integrity and stability of the mine tunnels,
water ingress, and the current locations and communication status of all underground mine personnel. In the aftermath of an accident, it can be
vital to maintain communications with trapped
miners and rescuers, and to establish and track
their positions. A knowledge of environmental
conditions through remote sensors in potential
escape routes would aid the preparation, planning, and execution of rescue operations.
Regardless of the specific type of high-stress
environment, reliable communication is essential
for successful mine operation under normal conditions, and is vital to the success of emergency
response and rescue operations. Communication
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
failures in hostile environments can occur
because of inadvertent destruction of network
infrastructure during normal operations as well
as in emergencies. Failures also occur due to an
error-prone communication channel in the particular application environment, and degradation
of the channel after an accident. The term communication channel here notionally lumps together all the physical components of the system
between the communicating devices, typically
including the material path through which signals must pass. Conventional communication
equipment may never be entirely adequate in
some severe high-stress environments, but it is
important to identify and investigate the various
characteristics that prevent satisfactory communication in such environments, both to map the
range of applicability of different approaches
and to indicate possible direction that may lead
to future advances.
This article provides a summary view of the
field, which may provide practical benefits to
other engineers who are working on similar
problems and projects. The remainder of the
article is arranged as follows. The next section
presents a general study of the channel characteristics for high-stress environments, and then
examines these characteristics in the specific
case of underground mines. We then outline the
various communication techniques used in
underground mines, and provide a brief survey
of the location sensing and tracking approaches
for these conditions. Wireless sensor networks
(WSNs) have recently been applied to this task,
and we provide a concise background and
overview of existing work on WSNs, and then
explore the design of WSN-based communication systems. This approach is supported by an
empirical study of the wireless communication
characteristics of typical commercial WSN nodes
deployed in an underground mine. The final section suggests potential research directions in the
field of underground mine communication and
concludes with a summary of the areas covered
in the article.
COMMUNICATION CHANNEL
CHARACTERISTICS
This section first outlines the general channel
properties that are common to what we have
called high-stress or harsh environments, and
then specializes them to the unique channel conditions prevalent in underground mines.
HIGH-STRESS ENVIRONMENTS
Communication channels in high-stress environments share several characteristics that make
reliable operation difficult [1–4].
Extreme Path Loss Due to Signal Absorption and Geometric Spreading — The transmitted signal is attenuated by absorption in the
medium through which the signal travels, and by
the geometric effect of the wavefront area
expanding as it propagates away from the transmitter. Both these effects cause a decrease in
signal strength with range from the transmitter.
The dependence on range typically has an
inverse power law with an exponent, which
depends on material properties of the medium
(which may vary with operating frequency and
environmental factors such as temperature and
humidity) and the geometry of the channel. The
absorption of electromagnetic (EM) waves in
water may be so high at usable frequencies that
acoustic links can be an alternative to radio or
optical links. EM signals are generally strongly
attenuated by the Earth at frequencies normally
used for wireless communications, but can penetrate large distances at ultra-low frequencies
(hertz to kilohertz).
IEEE
BEMaGS
Extensive Multipath Propagation and Fading — When a transmitted signal travels by multiple routes (i.e., multipath) to a receiver (e.g.,
by reflection from surfaces in the environment),
they get added at the receiver antenna. This sum
typically ranges between a maximum corresponding to the case when all the individual signals add in phase, to a minimum, even zero,
when the signals cancel. The random addition
causes space and time fluctuations in signal
strength which varies with receiver and transmitter position, signal frequency, and also movement of the transmitter, receiver, or reflecting
surfaces (which may be, e.g., vehicles). When
different paths have large length differences,
their corresponding signals interfere to cause
multipath fading and overlap in time, and may
result in distortions causing a degradation in the
link quality.
F
When different
paths have large
length differences,
their corresponding
signals interfere to
cause multipath
fading and overlap
in time, and may
result in distortions
causing a
degradation in the
link quality.
Rapidly Changing Time-Varying Channels —
Rapid motion of portable communication equipment, as well as variations in the intervening
channel caused by the motion, can cause
Doppler frequency shifts and rapid signal
strength fluctuations as multipath conditions
change. Underwater communication devices may
also encounter equivalent acoustic conditions as
a result of the motion of the ocean surface and
waves in internal water strata.
Large Propagation Delay and High Delay
Variance — This is a prime challenge faced by
very-long-range communication devices (e.g.,
satellites and deep space communication), as
well as underwater communications where the
acoustic propagation is some 200,000 times slower than EM waves in air. Variations in the effective path length of the signals due to
non-homogeneous material along the path can
cause changes to the total propagation time and
also introduce a large variance in path delays.
Noise — Noise in the communication system,
whether externally or internally generated,
reduces the effective system sensitivity and
therefore maximum range. Some environments
(e.g., the vicinity of high-power electrical motors)
can have high noise levels, which can degrade
radio communication. Noise levels can be severe
in satellite and deep space communications
because of EM radiation from transient solar
storms and background astronomical sources. In
the ocean storm, wave motion, shipping, and
even biological activity can generate severe
acoustic noise.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
67
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
attenuation. Restrictions on LOS communication arise from the normal mine arrangement of
long orthogonal tunnels, support pillars, tunnel
blockages, and floor undulations.
Miner 1
transmitting
Leaky feeder cable
Signals are leaking along the
entire length of the cable.
Ionized Air — Fires generate ionized air, which
can act as a plasma and disturb EM propagation
in mines.
Humid and Warm Conditions — The relative
humidity in mines is high, typically greater than
90 percent and the ambient temperature is commonly around 28°C.
Miner 2
receiving
Figure 1. Wireless communication mediated through fixed leaky feeder cables.
Besides these factors, stringent power constraints, topological variability, lack of interoperability, and the use of fixed communication
infrastructure [1] are important characteristics of
harsh environments. As a consequence of all
these factors, communication systems may suffer
from limited bandwidth, intermittent link connectivity, high distortion and link error and
packet loss rates, unacceptable packet reception
jitter, and delay (important in the case of lowlatency applications such as voice and video).
UNDERGROUND MINES
Underground mines are generally structurally
non-uniform, with a network of interconnected
tunnels, crosscuts, shafts, escape ways, first-aid
stations, alcoves, and tunnel blockages. Some of
the tunnels may contain rail tracks and conveyor
belts. The walls are generally rough and the
ground surface uneven, and scattered regions of
accumulated water may be present. Some parts
of the wall and ceilings may be strengthened
with bolted wooden grids and metal beams.
Environmental conditions that affect communication in mines include the following [1, 5].
Dynamic Changes in Underground Topology — The location of mine walls and faces may
alter continuously as a result of mining operations.
68
Communications
IEEE
Gaseous Hazards — The main component of
the flammable gases that leak from coal seams is
methane. When the concentration of methane
exceeds a critical threshold, an explosive mixture
is formed with a risk of gas blasts. Hence, continual ventilation is required to decrease the buildup of dangerous gases. However, in the case of a
disaster, power supply to the mine equipment
may be cut, possibly leading to failure of the ventilation system with the risk of dangerous gas
accumulation. Equipment for use in coal mines
in most jurisdictions must be certified as explosion-proof (i.e., unable to trigger an explosion in
air containing any proportion of methane).
Besides these natural environmental conditions, every mine is unique with its own distinct
operating considerations. In addition to the
above environmental properties, there are other
channel characteristics specific to underground
mines.
Waveguide Effect — Mine tunnels can act as
waveguides at certain frequencies, and allow relatively low-loss propagation, which can provide
long-range communication. This behavior is discussed in more detail in the next section.
Noise — The EM channel is effectively shared
with all the other electrical systems in the mine,
leading to background noise. Electric machinery,
power cabling and other mining appliances can
generate noise in some of the frequency bands
used by underground communication devices,
and hence can have an adverse effect on their
performance. Other independent systems using
wireless links can also contribute to the background noise. In a disaster response and recovery
situation, noise levels may be temporarily
reduced due to power shutdowns, but heavy
mechanical rescue equipment and other electronic equipment may introduce additional noise.
UNDERGROUND
MINE COMMUNICATION
Instability in Mine Structures — Some extraction techniques use collapse zones where there
are no supports and the faces are allowed to collapse as mining operations proceed, or in the
event of seismic activity.
This section describes some of the communication techniques that have been applied in underground mines, and outlines recent approaches to
communication and tracking devices.
Limited Line of Sight— Having a line of sight
(LOS) between transmitter and receiver can significantly improve communication, as signals can
propagate directly rather than through material
or around corners, both of which cause excess
Communication techniques applied in mines can
be classified as one of three basic types [5]:
Through-the-Wire (TTW), Through-the-Air
(TTA), and Through-the-Earth (TTE).
COMMUNICATION TECHNIQUES
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Advantages
A
BEMaGS
F
Name
Type
Disadvantages
Telephones
TTW
Easy operation
Vulnerable to damage from roof falls, mine fires, and explosions
Pager phones
TTW + TTA
Cheap; simple operation
One-way
Trolley
phones
TTW
Fixed/mobile — can provide communication to all rail haulage vehicles
Limited coverage; constant vibration; warm, humid, and dusty
conditions; interference from electrical machinery
Hoist
phones
TTW
Simple operation
Limited to communication between the hoist cage and surface/underground stations
Walkietalkie
TTA
Wireless communication; portable; twoway; can connect to nearby communication infrastructure (e.g., leaky feeder)
Generally poor range but may have good LOS performance
Table 1. Communication devices.
Through-the-Wire — As in systems deployed
above ground, a fixed infrastructure can provide routine long-distance communication in
harsh mine environments. Signals can be sent
over electrical conductors such as twisted pair
and coaxial cables, and via optical fibers [1, 2].
Cabling primarily intended for other purposes,
such as to provide power to electric rail vehicles (trolleys), can also be used to carry signals.
A major disadvantage of these systems is that
underground personnel must use equipment
that is physically connected to the cables for
signaling, whereas communication with unhindered mobility is a prime requirement underground. Hybrid systems, such as those using
leaky feeders, which use fixed wiring to distribute signals accessed by wireless connections
to nearby miners (Fig. 1) will be discussed in the
next section.
Although the performance of TTW systems
is satisfactory for routine operations, fixed
cabling is prone to damage and breakage in
accidents involving fire, earth falls, and tunnel
disruptions, and is difficult to maintain [5]. In
order to improve the reliability of TTW systems, various cable protection schemes have
been applied, including deployment through
conduit, burying the cable, feeding cables
through borehole connections to main lines,
and redundant cabling [1]. However, these
methods are expensive, make maintenance
more difficult, and increase system complexity.
Fiber optic cables have a significant advantage
over conventional wired communication techniques as they are not susceptible to electrical
interference and generally have far lower attenuation with distance. Some existing communication devices [2, 5] that use TTW techniques are
shown in Table 1.
Through-the-Air — TTA systems use wireless
links to allow untethered mobile communications. The environmental conditions in both metalliferous and coal mines present a unique set of
challenges for wireless communications. A simple model of a wireless communication system
comprises a transmitter, which generates and
launches an EM signal, the communication
channel through which the signal propagates,
and a receiver. Apart from the practical constraints on portable transmitter and receiver
design, the main difficulties in an underground
wireless system arise from the properties of the
communication channel and noise sources.
In general, EM propagation between two
arbitrary points in a mine level requires propagation through the Earth, down tunnels, around
corners and past machinery blockages. All these
conditions cause strong attenuation and signal
degradation, dependent on both operating frequency [1] and the specific environment.
The material through which mine tunnels are
constructed typically behaves as a low-loss dielectric, allowing a tunnel to act as a waveguide,
with relatively low-loss EM propagation possible
along it [5]. Ideal waveguides have a characteristic frequency called the cutoff frequency, below
which EM waves cannot propagate. The cutoff
frequency is directly related to the tunnel crosssection dimensions, and for typical mines the
cutoff frequency is in the tens to low hundreds
of megahertz (i.e., the very high frequency
[VHF] band). Above the cutoff frequency, EM
waves can propagate by essentially following
paths that bounce along the tunnel walls at a
grazing angle. At each reflection, some signal
energy is lost by scattering from irregularities in
the tunnel walls and floor, and refraction into
the surrounding material. The loss tends to be
greater at higher frequencies. LOS waveguide
propagation can be surprisingly good at ultra
high frequency (UHF) [6], with the best performance in coal mines typically at around 900
MHz in the UHF band and providing ranges of
some hundreds of meters. Below the cutoff frequency, waveguide propagation is not possible,
although direct LOS propagation can allow communication over a short range.
Long-range across-mine communications can
be implemented with hybrid systems in which
signals are carried sequentially in both fixed
TTW infrastructure and generally shorter wireless links. The TTW infrastructure can have
translation or bridge equipment at regular spacing to convert signals from cable form to wireless UHF signals, for example, which can be
used by miners with portable and handheld
UHF equipment. This approach combines some
of the benefits of both TTW and TTA systems,
but also carries the disadvantage that the fixed
infrastructure is vulnerable in mining accidents.
Two common hybrid forms use either UHF or
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
69
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Extremely low
frequency
(ELF)
(30–300 Hz)
Very low
frequency
(VLF)
(3–30 kHz)
Low
frequency
(LF)
(30–300 kHz)
Medium
frequency
(MF)
(300–3000 kHz)
Very high
frequency
(VHF)
(30–300 MHz)
Decreases
Bandwidth
Increases
Increases
Antenna
size
Decreases
Decreases
Attenuation
Increases
Increases
Noise level
Decreases
A
BEMaGS
F
Ultra high
frequency
(UHF)
(300–3000 MHz)
Figure 2. Influence of operating frequency.
medium frequency (MF) signals at around 1
MHz. A UHF implementation typically uses
leaky feeder cables mounted along selected tunnels as the fixed infrastructure. Leaky feeder
cable is specially constructed to allow a proportion of EM signals traveling in the cable to both
escape into the environment and enter the cable
from the environment. Two separated miners,
each near the leaky feeder, can communicate via
this cable using UHF handsets. The signal transmitted by one is picked up by the leaky feeder
cable, propagates down the cable while being
partly re-radiated into the environment along
the whole cable length, and the second miner
can receive this signal.
MF signals strongly couple to continuous
metal conductors and can use them as the longrange transmission medium rather than specially
constructed leaky feeders. Miners use MF equipment, generally bulkier and less portable than
UHF equipment, to generate signals that are
carried along purpose-deployed single metallic
conductors, or suitable pre-existing structures
such as lifelines or power rails. One benefit of
MF systems is that, in case of an accident, it may
be possible to use any available undamaged conductors to traverse blocked tunnel regions.
A natural extension of the hybrid approach
uses a deployment of wireless nodes to form a
wireless mesh network, which can forward messages from a miner within range of any node to a
destination point in the network where the message can be either delivered or forwarded through
other communications systems. The availability of
multiple paths in these networks gives them resistance to link failures, which are likely to occur in
emergencies. Digital modulation technologies
(e.g., as used in WiFi networks) have been developed to operate at high data rates in the severe
multipath environments typical of mines, and
hence may be able to support speech and video
communications. Because of their flexibility and
potential performance in a range of difficult environments, wireless mesh networks are the subject
of active current research.
70
Communications
IEEE
All the above systems can be combined to
extend the range over which communication is
possible and to provide system redundancy by
providing independent modes of operation.
Through-the-Earth — The attenuation of EM
signals through the ground strongly depends on
operating frequency. Figure 2 shows the qualitative dependence of several factors: available signal bandwidth, attenuation, antenna size, and
noise levels on the frequency. These trade-offs
prevent a system operating in one frequency
band from satisfying all operational and emergency requirements.
EM signals at the operating frequencies typically used by TTA systems are unable to penetrate rock strata. However, attenuation of EM
signals through the earth (TTE) decreases with
frequency, and at very low frequencies, ranges
can become great enough to allow even direct
surface-to-underground communication [5]. TTE
communication systems typically operate between
90 Hz and 4 kHz, and typically must use large
loop antennae to launch EM signals efficiently at
these frequencies. The data rates required for
speech cannot be supported at these frequencies,
so communications are limited to text messages.
Efficient antennas must be large (perhaps even
kilometers in diameter to support direct surfaceto-miner operation), and miners may have to
deploy wire loops underground as required. The
capability of direct communication with trapped
miners, independent of below-ground mine infrastructure, makes provision of TTE systems particularly important in mine emergencies.
TTW, TTA, and TTE communication technologies have their distinct capabilities and limitations, which makes selection of a suitable
system or combination of systems strongly
dependent on the particular application [2].
TRACKING SYSTEMS
The majority of current tracking systems are based
on radio frequency identification device (RFID)
technology. RFIDs or tags are small electronic
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
devices that can communicate with more complex
reader nodes via wireless communications. Reader
nodes can interrogate tags to exchange identity
and other information. Two main systems are currently being used. In the first, miners carry RFID
tags, and reader nodes placed at known fixed positions are connected to a mine communications
system, which can send data back to a central collection point. When a miner passes within range
of a reader, the tag identity is transmitted back to
the control center, giving an indication that a particular miner is located within the reception zone
of the reader. Position resolution depends on the
density of reader nodes in the mine and their coverage areas.
In the alternate system known as reverse
RFID, the RFID devices are deployed at known
positions. The miners now carry portable readers, which interrogate the static RFID devices
for their identity and transmit the information
back to a central site via the miners’ existing
communication system.
As mentioned previously, underground wireless communications can be implemented via a
mesh network of fixed nodes (Fig. 3). This raises
the possibility of integrating communication and
positioning functions in one system. At its simplest, the identities of the nodes nearest to the
miner can give zone position information. System performance can be improved by combining
link information, such as signal strength and
propagation timing measurements, to allow
more precise localization.
Information about research agencies, manufacturers, and commercially available tracking
systems can be found in [1, 7].
WSNS IN UNDERGROUND MINES
The general dependence on TTW systems for
operational use, together with restricted environmental monitoring capabilities, is a limitation in
providing safety assurance and rescue communication capabilities. This section investigates the
feasibility of applying the emerging WSN technology to implement a location sensing and environmental monitoring system, and discusses
related work and our own experiences in the
deployment of a WSN in an underground mine
in Parkes, New South Wales, Australia.
BACKGROUND
WSNs provide a new option for portable wireless communication systems, by using a network
of WSN nodes to provide the required network
connectivity in a cheap and efficient manner.
WSN devices are also well suited to distributed
environment monitoring, and can report gas and
dust concentrations and geological stability data
over their deployment range by attaching suitable sensors.
F
Figure 3. Node-based tracking system: a wireless mesh network.
for integration with deployed systems and
planned enhancements
• Modifications required to make WSN nodes
usable in the mining environment and able
to provide the desired data
• Long sensor node life through use of both
batteries high in energy density or rechargeable using available energy sources, and
techniques to minimize node power consumption while carrying out network operations
• Physical protection of the WSN nodes and
sensors to prevent damage or faulty operation in normal and post-accident circumstances without adversely affecting
communications
• Network protocols to store, exchange, and
retrieve information reliably under harsh
operating conditions
• System health monitoring to establish and
report the functional status of the system
during normal conditions as well as after
the occurrence of a mine accident
• System maintainability, that is, the effort
required to keep the system operational in
both normal and emergency conditions
• Decision systems to present sensor data in a
way that can be easily interpreted to assist
operational and emergency planning
AN EMPIRICAL STUDY IN AN
UNDERGROUND MINE
There are a number of important factors that
must be considered in order to design a WSN
implementation for location sensing and environmental monitoring in underground mines:
• Availability of sensors and nodes suitable
In order to assess the limitations of currently
available commercial WSN nodes when deployed
as a wireless communication network in underground mines, we conducted a series of experiments using off-the-shelf MicaZ [1] wireless
IEEE Communications Magazine • April 2010
IEEE
BEMaGS
The system reconfigures when a node in a route fails
and determines a new route for communication.
REQUIREMENTS FOR A
WSN IN HARSH ENVIRONMENTS
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
71
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
0
Base
station
10
8.5 m
9
1
14 m
12 m
2
8
8.8 m
12 m
3
7
11 m
4
9.5 m
4m
6
5
5m
A
BEMaGS
F
received by the base station from a majority of
the motes. Hence, a second experiment was conducted with the motes placed much closer to
each other, as shown in Fig. 4.
Figure 5 shows the percentage of packets
received correctly at the base station from each
of the individual motes. The success rate is less
than 50 percent for most of the motes. The
results suggest that those motes at one hop distance from the base station (motes 1 and 10)
and with a clear LOS (mote 10) performed better than the other motes. Several factors were
identified that could have contributed to the low
packet reception rate: dynamic channel changes
due to personnel motion during the tests; slight
misalignment of antennas due to the mounting
method, and multiple reflections from the mine
walls and other metallic objects attached to
them. In addition, we believe that variations in
the performance of the individual mote radios
also influenced packet throughput, as some
motes achieved good signal strength and performance, while others failed to communicate at
the same distance in a similar configuration. Our
experience highlights the need for custom design
of wireless sensor nodes that can provide reliable communication in harsh environments.
EXISTING WORK
Figure 4. Experimental setup inside the mine.
sensor nodes (motes) in an underground gold
and copper mine located near Parkes, New
South Wales, Australia.
Deployment — The mine tunnel that was accessible for experimentation was approximately 5 m
in width and 10 m in height, and had projecting
bolts positioned approximately 2 m above the
tunnel floor. MicaZ motes were enclosed in
plastic boxes to act as a protective casing, with
the antennas protruding. The plastic boxes containing the MicaZ motes could be mounted on
the bolts in the tunnel walls. Mote 0 was configured as the base station, while all the other
motes (numbered 1–10) were sited along both
walls of the tunnel, as depicted in Fig. 4. An
experiment was conducted to test whether motes
1–10 were able to successfully send packets to
the base station across multiple hops. All the
motes were programmed using TinyOS and nesC
[1]. The packet length was fixed at 29 bytes with
a simple structure comprising a header and a
payload containing the mote identification code.
Packets were sent at intervals of 100 ms, and
approximately 6000 packets were sent from each
mote over a period of 10 min.
Discussion — It proved to be more difficult
than expected to set up the experiment in the
humid and dusty mine environment. Coordinating the deployment of sensor motes inside a
dark underground mine tunnel and conducting
experiments is a nontrivial task in practice, as
the acoustic properties of the tunnel do not permit people to speak to each other if they are
more than 50 m apart. When the motes were
placed at 15 m intervals, no packets were
72
Communications
IEEE
We provide here a brief survey focusing on work
directed to the systems design and deployment
of WSNs in underground mines. Li et al. [8] present a sensor network deployment and collaborative communication strategy to detect the
structural changes in the event of underground
mine collapses. Field studies were conducted
through the deployment of a prototype system
consisting of 27 Crossbow Mica2 motes in the D.
L. coal mine in China. The prospects of using
ultra-wide-band (UWB) signals in conjunction
with WSNs for localization in underground
mines have been studied by Chehri et al. [9].
Measurement data for simulation were collected
from the CANMET experimental mine in Canada. Xuhui et al. [10] describe the implementation
of a methane gas sensor and propose an automatic calibration technique with the help of network connectivity. FireFly, a new sensor
hardware platform based on a cross-layer solution for tracking and voice communication in
harsh environments, was introduced in Mangharam et al. [11]. The experimental results
reported in that work were collected in a NIOSH
experimental coal mine. Xiaodong et al. [12]
describe their experiences in monitoring the coal
mine conditions via a wireless network consisting
of Crossbow MicaZ sensor nodes equipped with
custom developed multifunctional sensor boards.
RESEARCH DIRECTIONS AND
CONCLUSION
The performance of communication and tracking systems in underground mines has not been
as actively or extensively researched as contemporary surface-based systems. There are few
existing systems, and there is limited public
information regarding implementation details
and actual performance in mines. Ensuring safe-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
ty of mining personnel is one of the dominant
issues driving the development of these systems.
However, the difficult conditions in mines and
the lack of practical approaches have prevented
the development of a robust and generally applicable safety system. In view of experiences
learned from mining accidents, there is a need to
research the applicability of new technologies in
mine environments. Current systems lack the
capability of sensing and assessing information
that could help in predicting the risk of an accident. There is a need to engineer early warning
systems to bridge this gap because there is only a
very limited capability to limit the intensity and
impact of a disaster when it strikes.
Currently available tracking systems only register that a person is within a certain region or
zone. Research is needed into autonomous and
robust tracking systems capable of high spatial
resolution in real-time continuous tracking.
Wireless systems using spread-spectrum or UWB
radios that promise accurate positioning together with robust communication in strong-multipath environments, and software defined radios
able to adapt to dynamic propagation conditions
are other promising research areas that could
address some of the challenges posed by radio
propagation in mines.
This article outlines features common to a
range of high-stress environments and described
the factors that may affect communication in
deployments in these environments. It identifies
those factors that present the greatest challenges
to reliable communication in underground
mines. Properties of the underground wireless
channel, and the design and implementation of
current approaches to communication and tracking using this channel are also discussed. Finally,
the article discusses the emerging technology of
WSN deployments in harsh environments and its
applicability in underground mines, and we
describe our preliminary experiments to assess
the operation of WSNs in mines using generalpurpose commercial nodes.
REFERENCES
[1] P. Misra et al., “Safety Assurance and Rescue Communication Systems in High-stress Environments,” tech. rep.
UNSW-CSE-TR-0912, Univ. New South Wales, 2009.
[2] Niosh Office of Mine Safety and Health Research, “Tutorial on Wireless Communications and Electronic Tracking,” working draft, May 2009; http://www.msha.gov/
techsupp/PEDLocating/WirelessCommandTrack2009.pdf
_____________________________
[3] I. F. Akyildiz, D. Pompili, and T. Melodia, “ Underwater
Acoustic Sensor Networks: Research Challenges,” Ad
Hoc Net. J., 2005, pp. 257–79.
[4] I. F. Akyildiz et al., “Interplanetary Internet: State of the
Art and Research Challenges,” Comp. Net., vol. 43, no.
2, 2003, pp. 75–112.
[5] L.K. Bandyopadhyay, S. K. Chaulya, and P. K. Mishra,
Wireless Communication in Underground Mines,
Springer, 2010.
[6] A.G. Emslie, R.L. Lagace, and P.F. Strong, “Theory of the
Propagation of UHF Radio Waves in Coal Mine Tunnels,” IEEE Trans. Antennas Propagation, vol. AP-23,
no. 2, Mar. 1975.
[7] P. Misra, D. Ostry, and S. Jha, “Underground Mine
Communication and Tracking Systems: A Survey,” tech.
rep. UNSW-CSE-TR-0910, Univ. New South Wales, 2009.
[8] M. Li and Y. Liu, “ Underground Coal Mine Monitoring
with Wireless Sensor Networks,” ACM Trans. Sensor
Net., vol. 5, no. 2, 2009, pp. 1–29.
[9] A. Chehri, P. Fortier, and P. M. Tardif, “UWB-Based Sensor Networks for Localization in Mining Environments,”
Ad Hoc Net., vol. 7, no. 5, 2009, pp. 987–1000.
IEEE
BEMaGS
F
70
60
50
40
30
20
10
0
1
2
3
4
5
6
Node ID
7
8
9
10
Figure 5. Success rate vs. node ID.
[10] Z. Xuhui and W. Sunan, “Design of a Wireless Sensor
Network for Methane Monitoring System,” 6th IEEE
Int’l. Conf. Industrial Informatics, 2008, pp. 614–18.
[11] R. Mangharam, A. Rowe, and R. Rajkumar, “Firefly: A
Cross-Layer Platform for Real-Time Embedded Wireless
Networks,” Real-Time Sys., vol. 37, no. 3, 2007, pp.
183–231.
[12] X. Wang et al., “Deploying a Wireless Sensor Network
on the Coal Mines,” IEEE Int’l. Conf. Net., Sensing,
Control, 2007, pp. 324–28.
BIOGRAPHIES
PRASANT MISRA (pkmisra@cse.unsw.edu.au)
_______________ is a Ph.D. student in the Networks Research Laboratory (NRL), School of
Computer Science and Engineering, University of New
South Wales (UNSW), Sydney, Australia. His research interests have been in the area of wireless sensor networks,
network embedded systems, and wireless networks. He is a
recipient of the Australian Leadership Awards (ALA) scholarship, awarded by the Australian Agency for International
Development (AusAID), Government of Australia. He
received his B.E. (Hons) in computer science and engineering from Sambalpur University, India, in 2006, and worked
as a senior software engineer in Keane Inc., Bangalore,
India, 2006–2008.
SALIL KANHERE received his M.S. and Ph.D., both in electrical
engineering, from Drexel University, Philadelphia, Pennsylvania, in 2001 and 2003, respectively. He is currently a
senior lecturer in the School of Computer Science and
Engineering, UNSW. His current research interests include
participatory sensing, vehicular communication, and wireless mesh and sensor networks.
DIETHELM OSTRY [M] (diet.ostry@csiro.au)
___________ is a research scientist in the Wireless and Networking Technologies Laboratory, ICT Centre, CSIRO Australia. His recent research interests
have been in the areas of wireless networks, data network
traffic characterization, optical packet networks, and wireless sensor networks. He holds a B.Sc.(Hons) in physics
from the Australian National University and an M.Comp.Sc.
from the University of Newcastle, Australia.
SANJAY JHA is a professor and head of the Network Group
at the School of Computer Science and Engineering at
UNSW. He holds a Ph.D. degree from the University of
Technology, Sydney, Australia. His research activities cover
a wide range of topics in networking including wireless
sensor networks, ad hoc/community wireless networks,
resilience/quality of service (QoS) in IP networks, and
active/programmable networks. He has published over 100
articles in high-quality journals and conferences. He is the
principal author of the book Engineering Internet QoS and
a co-editor of the book Wireless Sensor Networks: A Systems Perspective. He is an Associate Editor of IEEE Transactions on Mobile Computing. He was a Member-at-Large,
Technical Committee on Computer Communications
(TCCC), IEEE Computer Society for a number of years. He
has served on program committees of several conferences.
He was the Technical Program Chair of the IEEE Local Computer Networks 2004 and ATNAC ‘04 conferences, and CoChair and General Chair of the Emnets-1 and Emnets-II
workshops, respectively. He was also the General Chair of
the ACM SenSys 2007 Symposium.
IEEE Communications Magazine • April 2010
Communications
A
80
Success rate (%)
Communications
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
73
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SERIES EDITORIAL
TOPICS IN INTEGRATED CIRCUITS
FOR COMMUNICATIONS
Charles Chien
I
Zhiwei Xu
Stephen Molloy
n the past ten years, we have witnessed rapid advances
in communication technology that enabled more than
an order magnitude increase in throughput and reduction
in power consumption. The steep rise in available throughput has stimulated the growth of ubiquitous broadband
services such as streaming of high-definition video contents, while on the other extreme, the steep drop in power
consumption, in particular for short-range radio technologies, has enabled personal area connectivity for all kinds of
portable consumer electronic devices, such as headsets,
cell phones, and cameras. Current third-generation (3G)
mobile devices easily support data rates on the order of
1–10 Mb/s, while quasi-static devices can easily support up
to 600 Mb/s with wireless LAN based on IEEE 802.11n.
For short distances, Bluetooth and ZigBee connectivity
technology consume power as low as 10–50 mW.
In the next few years, we will continue to see expansion
of available throughput to meet the increasing demand to
access high-definition (HD) contents such as Blu-ray quality video over the Internet. Such increase in demand has
propelled infrastructure upgrades for higher-speed backbone technology such as Data over Cable Service Interface
Specification (DOCSIS) 3.0, which achieves four to eight
times higher throughput compared to its predecessor, version 2.0. Typical deployment of DOCSIS 3.0 achieves
throughput of 171–343 Mb/s by means of channel bonding.
The improved throughput makes it possible to download
high-definition contents from the Internet with low latency. A further push in broadband capabilities will be driven
by enhanced visual renderings such as 3DTV, which captures stereo information in two optical polarizations, one
for each eye. At a minimum, the additional stereo information doubles the throughput requirement. While 3DTV
volume has reached only 1.2 million in 2009, its volume is
projected to hit 46 million by 2013.
Another future trend points to ultra-low-energy radio
systems for remote monitoring of vital signs in patients or
people with pre-existing health conditions. Such systems
require implants that make regular replacement of batteries inconvenient. Ideally, suitable technologies should be
capable of replenishing wasted energy by harvesting elec-
74
Communications
IEEE
trical energy from the environment (e.g., body motion,
ambient light, and thermal gradient). Existing short-range
radios such as ZigBee, adopted for building automation
and home energy management, still dissipate too much
power to self-sustain based on energy harvesting.
In this issue of the Topics in Circuits for Communications Series, we have selected three articles that mark
recent progress in the communications semiconductor
industry for highly integrated radio system-on-chip (SoC)
that enables future trends in broadband delivery of
enhanced high-definition video contents and self-powered
health monitoring systems.
In the first article, “Video Encoder Design for High
Definition 3D Video Communication Systems,” the
authors address the challenges to realizing efficient
encoders for emerging bandwidth constrained consumer
video applications, which have recently expanded beyond
current HDTV to the higher-resolution quad-HDTV as
well as three-dimensional video. Two key challenges in
reaching these incredible processing rates are memory
bandwidth and the complexity of context-adaptive binary
arithmetic coding (CABAC). The article describes new
architecture and circuit techniques to address these two
key challenges. Novel caching is applied to minimize external memory bandwidth, while algorithm parallelism at the
frame level is applied to the CABAC bottleneck. The
authors then demonstrate an example implementation and
test chip for the architecture, capable of encoding a single
view at a resolution of 4096 × 2160 or multiple views at
lower resolution.
The second article, “An Embedded 65nm CMOS Baseband IQ 48MHz-1GHz Dual Tuner for DOCSIS 3.0,”
exemplifies a fully integrated embedded complementary
metal oxide semiconductor (CMOS) digital dual tuner for
DOCSIS 3.0 and set-top box applications. To compete
with higher throughput offered by gigabit passive optical
network (GPON) and very high bit rate digital subscriber
line (VDSL), cable providers introduced DOCSIS 3.0,
which offers an increase in throughput by bonding multiple
downstream channels. Coexistence with high-powered analog cable transmissions imposes substantial challenges on
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
SERIES EDITORIAL
the integrated tuner design that has to demodulate eight
downstream channels simultaneously. To maintain low
power dissipation, the authors describe a cost-effective
multichannel and multituner solution by integrating the
tuner, digital demodulator, MPEG decoder, memory, and
processor core on a single SoC. The article covers several
critical design issues, such as harmonic rejection for
demodulation, image rejection with imbalance canceller,
and high dynamic range front-end. The authors also provide detailed design trade-offs in the tuner architecture
and RF circuit design issues, as well as the benefit of their
unique implementation with respect to power consumption
and cost in future technology scaling.
The third article, “Integrated Electronic System Design
for Implantable Wireless Batteryless Blood Pressure Sensing Microsystem,” reviews the recent techniques used in
real-time monitoring of blood pressure to identify genetic
susceptibility to diseases. These techniques provide critical
research tools to develop new treatments for cardiovascular and hypertension. In contrast to conventional monitoring techniques that rely on invasive bulky catheter tip
transducers, an implantable miniature lightweight blood
pressure sensing microsystem with wireless data communication and adaptive RF powering capability is highly desirable. However, the integration of a batteryless wireless
communication system with real-time blood pressure monitoring remains a challenge. The authors describe design
challenges with respect to various circuits and microsystem
impairments such as continuously changing RF power coupling and magnetic field. The article then describes design
techniques to boost the circuit immunity to dynamic environment and interferences, and demonstrate the integrated
blood pressure sensing microsystem for future implantable
integrated health monitoring systems.
We would like to take this opportunity to thank all the
authors and reviewers for their contributions to this series.
Future issues of this series will continue to cover circuit
technologies that are enabling new and emerging communication systems. If the reader is interested in submitting a
paper to this series, please send your paper title and an
abstract to any of the Series Editors for consideration.
BIOGRAPHIES
CHARLES CHIEN is president and CTO of CreoNex Systems, which focuses on
technology development for next-generation systems. Previously he held
various key roles at Conexant Systems, SST Communications, and Rockwell. In his career he has architected several key products including a
CMOS/SiGe chipset for multimedia over coax (MoCA), an IEEE 802.11abg
WLAN RF CMOS transceiver and GaAs PA/RF switches, a wireless audio
CMOS chipset for home theatre in a box, CDMA2000 cellular RF CMOS
transceivers, and low-power wireless networked sensors. He was also an
assistant adjunct professor at the University of California at Los Angeles
(UCLA) from 1998 to 2009. His interests focus mainly on the design of
system-on-chip solutions for wireless multimedia and networking applications. He has published in various journals and conferences, and has
authored a book entitled Digital Radio Systems on a Chip. He received his
B.S.E.E. from the University of Caifornia at Berkeley, and his M.S. and
Ph.D. from UCLA. He was a member of the technical program committee
of ISSCC from 1998 to 2006.
ZHIWEI XU received B.S. and M.S. degrees from Fudan University, Shanghai,
China, and a Ph.D. from UCLA, all in electrical engineering. He held industry positions with G-Plus Inc., SST communications, Conexant Systems, and
NXP Inc., where he did development for wireless LAN and SoC solutions for
proprietary wireless multimedia systems, CMOS cellular transceivers, MoCA
systems, and TV tuners. He is currently with SST as department head, working on various aspects of wireless communication SoC and software
defined radios. His current research interests include wireless communication SoCs for high data throughput as well as ultra-low-power applications.
He has published in various journals and conferences, made one contribution to the Encyclopedia of Wireless and Mobile Communications, and has
four granted and five pending patents.
STEPHEN MOLLOY received M.S. and Ph.D. degrees in electrical engineering
from UCLA in 1993 and 1997, respectively, where his research focused on
low-power circuits and architectures for video signal processing. This work
led to the award of the Showman Prize from UCLA in 1997, and resulted in
over a dozen conference and journal publications. He received a B.S.
degree in electrical engineering from Rensselaer Polytechnic Institute in
1991. He served as Associate Editor of the IEEE Journal of Solid-State Circuits from 2001 to 2004 and was a member of the technical program committee of the IEEE International Solid-State Circuits Conference from 1998
until 2005. He is currently vice president of engineering at Qualcomm,
leading architecture development.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
75
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
INTEGRATED CIRCUITS FOR COMMUNICATIONS
Video Encoder Design for
High-Definition 3D Video
Communication Systems
Pei-Kuei Tsung, Li-Fu Ding, Wei-Yin Chen, Tzu-Der Chuang, Yu-Han Chen, Pai-Heng Hsiao, Shao-Yi Chien,
and Liang-Gee Chen, National Taiwan University
ABSTRACT
VLSI realization of video compression is the
key to real-time high-definition 3D communication systems. The newly established multiview
video coding standard, as an extension profile of
H.264/AVC, draws more and more attention for
its high compression ratio and free-viewpoint
support. Besides providing the 3D experience,
multiview video can also give users complete
scene perception. However, the multiple-viewpoint throughput requirement of MVC increase
the complexity and hardware cost dramatically.
The system memory bandwidth, on-chip memory
size, and processing data throughput of each
module all need to be optimized in an MVC
encoder. Therefore, efficient hardware solutions
for MVC architecture design are needed. In this
article an overview of 3D video coding standards
developments and design challenges of an MVC
encoder are discussed. Then the algorithm and
architecture optimization schemes are proposed.
For the trade-off between system memory bandwidth and on-chip memory size, a cache-based
prediction engine is proposed to ease both
design challenges. Moreover, the hybrid openclose loop intra prediction scheme and the
frame-parallel pipeline-doubled dual CABAC
solve the throughput requirement problem. At
the end of this article, based on all the proposed
solutions, a prototype single-chip MVC encoder
design with processing ability of 4096 × 2160 single-view to 1280 × 720 seven-view is presented.
INTRODUCTION
For advanced TV applications, vivid perception
quality is required. Therefore, higher and higher
video resolutions, like high-definition (HD) 720p
(1280 × 720 pixels) and 1080p (1920 × 1080 pixels), are recommended. In addition, 3D video
can bring the 3D and realistic perceptual experience to viewers by projecting different views to
users’ left and right eyes simultaneously. As the
technology evolves, lots of 3D related applications, such as 3D-TV and free-viewpoint TV, are
emerging [1].
76
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
In a real-time HD 3D video communication
system, three key technologies make it feasible.
The first one is the stereo or multiview capturing
and display device. The second one is the coding
standard. Since 3D video contains many different
view angles, different and more efficient coding
algorithms than the conventional single-view
video coding standards are required to further
reduce the bit rate for communication. Third, the
efficient hardware architecture is required for
accelerating the coding speed to meet the realtime constraint. Because of the multiple-viewangle characteristic, data needed to be processed
in a 3D video is multiple times that in a conventional single-view video. Thus, if the conventional
architecture is adopted, it will multiply computation complexity and hardware cost.
In order to transmit and store 3D/multiview
contents, an efficient multiview video coding
(MVC) scheme is needed. The MPEG 3D
Audio/Video (3DAV) Group is working on the
standardization of MVC. In July 2008, MVC was
standardized as the Multiview High Profile in
H.264/AVC by the MPEG 3DAV Group [2].
The joint MVC (JMVC) was released by the
MPEG 3DAV Group as the reference software
and research platform [3]. In the JMVC
H.264/AVC is adopted as the base layer. In
addition, disparity estimation (DE) and disparity
compensation (DC), the most significant features in JMVC, can effectively discover the
interview redundancy of a multiview video and
save 20–30 percent of bit rates. Based on the bit
rate reduction, an HD MVC sequence is able to
be stored in high-end multimedia portable storage like a Blu-ray disc. However, the coding
complexity increases dramatically in the MVC
because of the hybrid inter-view DE and intraview motion estimation (ME) prediction
schemes. Furthermore, the processing throughput requirement of HD MVC is many times
larger than that of the current HDTV specification. Thus, a new and efficient encoder architecture design for the MVC is desired. In this article
the mainstream 3D video coding standards,
design challenges in MVC encoder design, and
the proposed solutions are briefly introduced.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
The video coding standard development is introduced in the next section. The hardware resource
analysis is then presented. Then the proposed
MVC architecture design is shown. The final
section concludes this article.
FROM 2D TO 3D: VIDEO CODING
STANDARD DEVELOPMENT
2D VIDEO CODING: FROM MPEG-1,
H.261 TO H.264/AVC
Video data without compression is impossible to
transmit directly due to the incredible size of the
uncompressed raw data. Since 1990 many video
coding standards have been defined for storage
and transmission. Among these coding standards, coding efficiency is the most important
criterion. There are two main series of video
coding standards: the International Standards
Organization (ISO) MPEG-x standards and
International Telecommunication Union —
Telecommunication Standardization Sector
(ITU-T) H.26x standards. The MPEG-x series
contains MPEG-1, MPEG-2, and MPEG-4. On
the other side, the H.26x series starts from H.261
in 1990 to H.263, H.263+, and H.26L. Furthermore, some standards are the result of coworking of these two groups. For example, MPEG-2
is also called H.262 and is the result of a common project. Then H.264 is delivered by both
ISO and ITU-T, which is also called the Joint
Video Team (JVT). Therefore, H.264 can also
be called MPEG-4 Advanced Video Coding
(AVC) or H.264/AVC. Being the latest finalized
advanced video coding standard from these two
main streams, H.264/AVC has the best coding
performance. It provides more than 50 percent
bit rate reduction over the previous MPEG-2
standard. In order to provide better and better
rate distortion (R-D) performance in the future,
at the last MPEG meeting, a new Joint Collaborative Team between MPEG and ITU was created to work on a new standard.
In order to solve the problems of the previous standards, MVC is proposed as an extension
profile of H.264/AVC. In contrast to the singleview-plus-depth format, MVC encodes video
data from multiple viewing angles into a single
bitstream by hybrid motion and disparity compensated prediction. Figure 1 illustrates the
overview of an MVC system and the corresponding block diagram of an MVC encoder. The
multiview video is captured by a camera array,
followed by the MVC encoder compressing the
multiview video data for transmission or storage.
On the decoder side, reconstructed multiview
video can be displayed on various displays such
as currently commercialized HDTV, or nearly
developed stereo and multiview 3DTV. In an
MVC encoder, video frames from the first view
channel are compressed by a typical H.264/AVC
encoder. On the other hand, DE and DC are
adopted to other view channels to further reduce
inter-view redundancy. This multiple-viewpoint
characteristic of MVC avoids the quality degradation from the inaccurate depth map. Furthermore, the H.264/AVC-based encoding flow
reduces the bit rate overhead for each view.
However, the complexity of an MVC encoder is
also much higher than that of the single H.264/
AVC encoder due to its multichannel characteristic. Therefore, an efficient hardware architecture is urgently required.
F
MVC outperforms
previous 3D video
coding standards
by use of
H.264/AVC-based
coding scheme. The
multiple view angles
characteristic also
avoids the quality
uncertainty due to
the depth map.
However, these
features also bring
the larger complexity
and hardware cost
than previous
standards.
MVC outperforms previous 3D video coding
standards by use of an H.264/AVC-based coding
scheme. The multiple view angles characteristic
also avoids the quality uncertainty due to the
depth map. However, these features also bring
larger complexity and hardware cost than previous standards, especially when the resolution
requirement is as high as the HDTV specifications. The main design challenges of an MVC
encoder are shown in Fig. 2 and discussed below.
ULTRA HIGH COMPUTATION COMPLEXITY AND
THROUGHPUT REQUIREMENT
3D video has always played an important role in
the video processing research field, including, of
course, 3D video coding. The first finalized 3D
video coding standard was the MPEG-2 Multiview Profile. A stereo video sequence can be
compressed into a bitstream containing a base
layer and an enhancement layer. In addition to
the stereo-view representation, another approach
to 3D video is the single-view-plus-depth, or socalled 2D + Z, format. The Advanced ThreeDimensional Television System Technologies
(ATTEST) from European Information Society
Technologies (IST) and MPEG-C Part 3 from
MPEG both focus on this format. The depth
information can be captured by the depth sensor. With the depth map, virtual views can be
generated by depth image-based rendering
(DIBR). However, the technology of depth map
generation is not mature enough. It directly
causes quality degradation of the rendered virtual views on the receiver side.
MVC has large computational requirements
because it needs to compress data from multiple
viewpoints. In a video coding system, inter-frame
redundancy elimination causes most of the complexity. For a single-view video, ME is used to
find out the inter-frame relationship and reduce
the data redundancy in the temporal domain. In
MVC DE is used as well for the inter-view
domain inter-frame prediction. For an N-view
multiview sequence, this hybrid ME/DE encoding
scheme requires more than N times more computation than a single-view sequence. Figure 2a
shows the integer ME/DE (IMDE) computation
analysis under different resolutions and view
numbers, where different search algorithms used
in integer ME/DE and the corresponding computation requirements are listed. Two hardware oriented algorithms are considered in Fig. 2a. The
full search algorithm uses all possible candidates
over the entire search window (SW) and thus
provides the best rate-distortion (R-D) perfor-
IEEE Communications Magazine • April 2010
IEEE
BEMaGS
DESIGN CHALLENGES OF AN
HD MVC ENCODER
3D VIDEO CODING:
FROM MPEG-2 MULTIVIEW PROFILE TO MVC
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
77
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
In the architecture
design field, the
A
BEMaGS
F
Output
various representation
Input
multiview video
system memory
bandwidth and the
HDTV
on-chip memory size
are two major
Stereo TV
limitations. The
trade-off between
MVC
encoder
them is classic in
architecture design.
Storage/
transmission
MVC
decoder
That is, larger
Multiview
3DTV
on-chip memory
allows lower system
memory bandwidth.
Input frame
From view 1
Block engine
Entropy
coding
Compressed
data
Intra prediction
Motion
estimation
Motion
compensation
Frame(s)
memory
MV
First view channel
MV and DV
Second view channel
Motion/
disparity
estimation
Vector
coding
Vector
coding
Frame(s)
memory
Motion/disparity
compensation
Intra prediction
Input frame
From view 2
Block engine
M
u
l
t
i
p
l
e
x
e
r
Bitstream
Entropy
coding
Third view channel
Nth view channel
Figure 1. Overview of an MVC system and the block diagram of an MVC encoder.
mance. However, it requires huge computation.
Hierarchical search is a fast algorithm to reduce
the computation. By hierarchically downsampling
the SW, the required number of search candidates can be reduced to about 10 times less than
that of full search. However, the computation is
still too large to be processed for the HD MVC
specifications. As shown in Fig. 2a, the required
instructions per second (IPS) is over 1000 GIPS
even when hierarchical search is adopted. Meanwhile, the high-end quad-core CPU by Intel,
QX9770, can only provide 60 GIPS. According to
this analysis, hardware acceleration is needed for
an HD MVC encoder design.
78
Communications
IEEE
Another design challenge of an HD MVC
encoder is the large data throughput requirement. To encode an N-view MVC sequence,
the throughput requirement is about N ore
more times that of encoding a conventional
single-view sequence. However, throughput on
some modules cannot be enlarged by simply
duplicating and parallel processing. Taking the
entropy coding, for example, the entropy coder
in H.264/AVC and MVC, content-based adaptive binary arithmetic coding (CABAC), has
very strong data dependence since it needs to
consider the previous symbols when generating
the current symbol. Therefore, most existing
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
CABAC symbol rate
2000
IMDE computation in MVC
1E+5
Maximum symbol count
Average symbol count
1800
Symbols/MB
Instructons/s (GIPS)
1600
1E+4
1E+3
60 GIPS==
QX9770
1E+2
1200
1000
800
600
10
Full search
Hierarchical search
Throughput limit of one-symbol CABAC
400
1
200
20
0
8
12
x7
x2
0
8
12
20
x7
x3
20
0
8
12
x7
(a)
x4
0
92
80
x2
0
x1
80
0
92
1
x3
0
0
x1
1 2 3
4 5 6 7 8
9 10 11 12 13 14 15 16 17 18 19 20
Frame number
1
(b)
SRAM size in MVC
System bandwidth in MVC
10000
10
Level C
Level C+
Level D
Hier.
1000
kbytes
Gbyte/s
1400
1
100
Level C
Level C+
Level D
Hier.
10
1
0.1
80
12
x
0
72
x2
80
12
x
0
72
x3
80
12
(c)
x
0
72
0
2
19
x4
x1
0
08
x2
0
08
0
2
19
x1
x3
20
80
12
x2
20
x7
80
12
x7
x3
80
12
(d)
20
x4
x7
20
19
80
x2
80
0
x1
20
x3
0
x1
19
Figure 2. Design challenges in an HD MVC encoder: (a) IMDE computation analysis; (b) CABAC throughput analysis; (c) system bandwidth analysis; (d) on-hip SRAM size analysis
CABAC coder designs can only provide the
throughput of one symbol per clock cycle.
However, this processing ability is far from the
target HD MVC throughput. Figure 2b illustrates the frame-by-frame symbol count analysis result on an HDTV sequence. The red line
is the largest throughput of a one-symbol
CABAC coder. This throughput limit is calculated from the operating frequency and video
resolution. Take our target HD MVC specifications, for example. Considering the systemon-chip (SoC) integration compatibility, the
highest operating frequency of the previous
H.264/AVC encoders is selected as no more
than 200 MHz [4–6]. However, when the target
specifications are as high as the HD MVC, the
available processing cycles for a macroblock
(MB) is only about 350 cycles even the operating frequency is increased to 300 MHz. As
shown in Fig. 2b, the symbol count has large
variance between frames because the symbol
counts of the I-frame and P-frame are much
higher than that of the B-frame. A conventional CABAC coder can barely deal with the average case, but is infeasible for the maximum
symbol rate. Unfortunately, the symbol rate
cannot be raised by increasing the parallelism
because of the data dependence issues mentioned above. Therefore, a new and efficient
architecture is required.
HIGH SYSTEM MEMORY BANDWIDTH AND
LARGE ON-CHIP MEMORY SIZE
Since hardware acceleration is needed according
to the analysis above, further system and memory analysis is required before the implementation. In the architecture design the system
memory bandwidth and on-chip memory size are
two major limitations. The trade-off between
them is classic in architecture design. That is,
larger on-chip memory allows lower system
memory bandwidth.
In a video encoder design, IME, or IMDE in
MVC, requires most of the bandwidth and onchip memory because a large SW must be loaded onto the chip for doing IMDE. Typically the
width and height of the SW are set to about 10
percent of the frame width and height, respectively. Furthermore, more than one SW is loaded when the frame type is B-frame or the
multiple-reference-frame scheme is enabled. In
order to reduce the hardware cost, various data
reuse schemes, including level C, level C+, level
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
79
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
350 Cycles/stage
Stage 2: IMDE
Stage 3: NOP
Stage 4: FMDE Pf
Stage 1: IMDE Pf
Stage 5:
FMDE
Stage 6:
IP and
MDC
Stage 7:
REC
Stage 8:
Dual-EC and DB
MVC encoder chip
View-parallel MB-interleaved CTRL
EC select
View1 cache SRAM
EC
core 1
IP
core
IMDE
prefetch
IMDE
core
Residue
MB SRAM
Bitstream buf.
Cur. MB
buf.
Cur. Luma
MB buf.
FMDE
core
FMDE
prefetch
Cur. Luma
MB buf.
EC
core 2
IMDE
core
MDC
core
Bitstream buf.
Cur. MB
buf.
IMDE
core
Rec. MB
SRAM
MDC MB
SRAM
DB MB SRAM
View2 cache SRAM
128-bit system bus interface
System external
memory
DRAM controller
IMDE: Integer ME/DE
FMDE: Fractional ME/DE
Bus mater/slave
IP: Intra prediction
MDC: Motion/disparity compensation
Processor
Video input
REC: Reconstruction
Pf: Prefetch
External
bus
EC: Entropy coding
DB: Deblocking
Figure 3. Proposed eight-stage MB pipelined MVC encoder architecture. Note that each stage has about 350 processing cycles if the processing frequency is 300 MHz under HD MVC specifications.
D, and hierarchical search, have been proposed
in recent years [7]. The system memory analysis
of these algorithms for MVC with different
numbers of views and resolutions are shown in
Figs. 2c and 2d. Different trade-offs between
bandwidth and memory size are selected under
different algorithms. For example, level D data
reuse has the largest on-chip SRAM size and
lowest memory bandwidth. From the bandwidth
point of view, a high-end SoC with a fairly wide
128-bit bus can only support about 4 Gbytes/s
bandwidth even under 100 percent bus utilization and 250 MHz operating frequency. Meanwhile, the required bandwidth is over 5 Gbytes/s
for nearly all algorithms listed in Fig. 2c for 1a
080p three-view MVC sequence. On the other
hand, if the TSMC 90LP process is used, the
lowest point in Fig 2d, which is about 60 kbytes,
occupies the equivalent gate count from 0.57 to
1.94 million under different memory compiler
80
Communications
IEEE
configurations. From Fig. 2d, when the target
specification matures, the maximum memory
requirement may be as high as dozens or even
hundreds of millions of gates, which is far
beyond what a high-end SoC system can support.
Therefore, a smart strategy to reduce both onchip memory size and system memory bandwidth
is desired.
PROPOSED MVC
ENCODER SOLUTIONS
SYSTEM ARCHITECTURE
Figure 3 shows the system architecture of the
proposed MVC encoder. The encoder contains
seven kinds of computation cores, including integer ME/DE (IMDE), fractional ME/DE
(FMDE), intra prediction (IP), motion and disparity compensation (MDC), reconstruction
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
(REC), entropy coding (EC), and deblocking filter (DB).
According to the design challenges described
above, instead of simply raising the parallelism
from the conventional three-or four-stage MB
pipelined architecture in the previous
H.264/AVC encoder design [4–6], an eight-stage
MB pipelining is proposed. In order to ease the
hardware cost of IMDE, the inter-frame prediction part is split into five MB pipeline stages,
and the cache-based prediction core is adopted.
After the cache memory is used, two SW
prefetching stages for IMDE and FMDE are
added to load SWs into on-chip SRAM prior to
the processing stage. They not only reduce the
burden of the pipeline-cycle budget but also
enhance the hardware utilization of IMDE and
FMDE cores. An no operation (NOP) stage is
inserted to deal with the data dependence
between the prefetching and processing stages.
After the inter-frame prediction is done, the
intra-frame prediction and motion/disparity compensation are performed in parallel in the sixth
stage. The reconstruction stage reconstructs the
compressed frame as the reference for the following frames. Finally, two CABAC EC modules
and one DB module are processed simultaneously in the eighth MB pipeline stage. According
to the analysis in the previous section, each
pipeline stage only has about 350 cycles under
the target HD MVC specifications.
PREDICTOR-CENTERED CACHE-BASED
MOTION/DISPARITY ESTIMATION
According to the analysis in the previous section,
the IMDE part accounts for most of the hardware cost. One major reason is that it requires a
large SW buffer, which grows proportionally to
the frame resolution. Based on previous work on
fast search, only 30 percent of the SW area is
really used in common intermediate format
(CIF), and this utilization decreases to 15 percent in D1 video. That is, much data is loaded to
the on-chip SW buffer unnecessarily. However,
if we directly shrink the search range, the R-D
performance drops greatly. These two characteristics indicate that we only need a small part of
data in the SW, but we cannot assume that the
location of this part is always close to zero-MV.
Therefore, a predictor-centered cache-based
IMDE is proposed. The SW is centered by the
predictor, so the search range can be reduced
with little quality degradation. The cache memory trades off the possibility of cache misses for a
much smaller on-chip memory capacity, and is
still able to handle the varying and dynamic data
access pattern.
Figures 4a and 4b are the comparison
between the conventional ME algorithm and the
proposed predictor-centered algorithm. Figure
4a shows the concept of previous hardware-oriented algorithms. In order to find the relationship between frames, a SW is set on the
reference frame around the relative location of
the current MB. That is, the center of the SW is
the zero motion vector (MV). Since the length
of MV grows proportionally to the dimension of
video frames, the size of SW also needs to be
enlarged to keep the best-matching MV inside,
or the quality drops greatly. To prevent this
from raising SW cost, the proposed algorithm
shown in Fig. 4b takes the relationship between
MVs into consideration. Since MBs inside the
same object should have similar MVs, MVs from
the neighboring MBs can be set as the initial
search hint of the current ME process. If we put
the SW around the best hint instead of the zero
MV, the required SW size can be dramatically
reduced because of the inter-MB MV similarity.
Based on this concept, the detailed algorithm
flow of the proposed predictor-centered algorithm is described as follows. First, several initial
hints are set, and each has a tiny SW. The window size is 4 × 4 in our implementation. Second,
each candidate in these windows is sent to the
IMDE module, and a corresponding R-D cost is
calculated. The candidate with the best R-D cost
is chosen as the refinement center, and a larger
refinement range is defined around it. However,
this multiple hints and refinement flow may
cause a larger quality drop in cases with non-uniform motion fields. A motion information preserving scheme is proposed to maintain the
quality on the complex motion field by getting
more accurate initial hints and refining centers.
In the proposed scheme motion information is
saved and reused in the intra-coded MBs. The
MV predictor defined in the standard
H.264/AVC is derived from the MV field. As a
result, when an MB is intra-coded, its motion
information is not encoded, and no MV is available. However, if the MV pointing to the best
matched block is stored, even if the intra mode
wins the inter/intra mode decision, the MV can
still be used as a hint for neighbor MBs. Therefore, motion information is reused instead of
being discarded even if the block is intra-coded.
After the proposed scheme, the R-D performance on all the test sequences used in JVT
H.264/AVC meetings can be maintained as less
than 0.1 dB drop even when the SW size is as
small as ±16 × ±16 under our target HD MVC
specifications.
Based on this multiple hints with refinement
scheme, the SW can be retargeted MB by MB
dynamically, and therefore the requirement on
SW size is reduced. Figures 4c and 4d illustrates
how the predictors are generated. The performance of this predictor-centered algorithm highly
depends on the accuracy of hints. If the hint targets a wrong region, it needs a larger refinement
range to compensate for the quality loss, and the
benefit of the predictor-centered algorithm is
decreased. Two kinds of hints are used to exploit
the spatial and temporal correlation of MVs inside
the same object. The first is the intra-frame predictors, which are MVs/DVs from the neighborhood MBs. Since the video processing is done in
raster-scan order, only MBs above or left to the
current MB have MVs available. Thus, the
MVs/DVs from the top, top-left, top-right, and
left MB are used as the intra-frame predictors.
Furthermore, the zero MV and motion vector predictor (MVP) defined in H.264/AVC and MVC,
which is the median-filtered result of the top, topright, and left MVs, are also allocated as intraframe predictors. The inter-frame predictors are
the other kind of predictors. Because an MVC
sequence consists of more than one viewpoint, one
IEEE
BEMaGS
F
A motion
information preserving scheme is
proposed to maintain the quality on
the complex motion
field by getting more
accurate initial hints
and refining centers.
In the proposed
scheme, motion
information is saved
and reused in the
intra-coded MBs.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
81
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
Because of the
splitting of the intra
Search
window
Frame
Current
macroblock
Initial search
hint
prediction and
reconstruction
stages, only one best
matched mode is
Ref.
frame
required for
Current
frame
Ref.
frame
Current
frame
reconstruction.
Consequently, the
ME
ME
cycle budget is
enough even under
scheme.
Time
Time
the close-loop
t
t–1
t–1
(a)
t
(b)
View 1
t
MV2
Time
Cur.
MB
MV1
DV2
Currrent frame
MV1 + DV1 ~=MV2 + DV2
MV2 from best matching MB
(c)
View
DV1
t–1
MVs from
neighboring
MBs
Current frame
View 2
.
(d)
Figure 4. Proposed IMDE algorithm: a) concept of previous hardware oriented algorithms; b) the proposed
predictor-centered algorithm; c) intra-frame predictor generation and reuse; d) inter-view predictor generation and reuse.
object may be captured in more than one view at
the same time. Since the object is the same, the
captured motion in different cameras are also similar. Therefore, the MVs from the neighboring
views are very strong predictors [8]. In fact, after
including the inter-frame predictors, the required
refinement range can be shrunk to 4 × 4, the same
size as the tiny search window for a hint. That is,
the refinement step in ME can be canceled for
those views under both ME and DE [9].
In order to support the dynamic hint refinement
access pattern without loading all the pixels in all
possible locations of SW, a cache system is implemented as the SW buffer. Unlike the conventional
cache memory system in the computer architecture
field, cache memory used in video processing has
several different features. The most significant difference between them is that video data has 2D spatial coherence rather than the 1D addressing in
general cache memory design. To fully utilize this
coherence, the internal index wraps in two dimensions. The three-tuple vector (x, y, frame-index) is
translated to the tag address and the tag. A tag set is
pointed by the tag address, and the tag is compared
to that set. Upon a cache hit, the word address
locates the word in a five-banked on-chip SRAM.
82
Communications
IEEE
The cache system provides flexible data access.
However, the cache miss penalty is considerable.
Every time the wanted data are not in the cache,
the system needs to be stalled, and the required
data is reloaded from the external memory. This
stall-and-reload waiting time lowers the hardware
utilization. Therefore, two new MB pipelines,
IMDE prefetch and FMDE prefetch, are added
to the proposed MVC system architecture to
lower the cache miss rate. After this scheduling
optimization and other proposed cache architecture optimizations, including priority-based
replacement policy and a concurrent SW
prefetching and reading scheme, the total cycles
of cache miss penalty are reduced by 93 percent.
That is, only 1.2 misses will happen during one
MB pipeline stage, which has 350 cycles.
HYBRID OPEN-CLOSED LOOP INTRA PREDICTION
Other than the inter-frame prediction, intraframe prediction is also used for reducing the
spatial redundancy within a frame. Pixels are
predicted from the neighboring pixels. In the
H.264/AVC high profile and MVC, there are
three kinds of intra predictions: intra4 × 4 (I4)
mode, intra8 × 8 (I8) mode, and intra16 × 16
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
D1 SDTV
1080p HDTV
Task
Setup
Intra_4x4 14 Intra_4x4 14
Blk0
Blk0
Blk0
Blk1
Pred.
Rec
Pred.
rec
I16
Blk0
...
Intra_4x4
14
Blk15
Blk15
Pred.
R ec
I16
Blk1 Chroma ... Chroma
Mode0
Mode4
5
I16
Blk0
...
Intra_8x8
910
Cycles
1300
(a)
Use reconstructed pixels
Blk
0
Blk
5
Blk
2
Blk
3
Blk
6
Blk
7
Blk
8
Blk
9
Blk
10
Blk
11
Blk
12
Blk
13
Blk
14
Blk
15
Use original
pixels
Prediction
stage
Setup
Blk
4
Blk
1
HD MVC
Task
Intra_16 Intra_16
Blk0/1
Blk2/3
Pred.
Pred.
...
Intra_4x4 Intra_4x4
Blk0/1
Blk2/3
Pred.
Pred.
...
Intra_8x8
Blk0
Pred.
Reconstruction
stage
...
Intra_16
Blk14/15
Pred.
Intra_4x4
Blk14/15
Pred.
Intra_8x8
Blk4
Pred.
Reconstruction of
the best pred. mode
from I16/I4/I8
Cycles
272
(b)
320
350
(c)
Figure 5. Issues and solutions on the intra stage: a) illustration of the throughput bottleneck due to data dependence; b) the pro-posed
hybrid open-close loop intra prediction; c) the corresponding processing scheduling.
(I16) mode. The 4 × 4 discrete cosine transform
(DCT) is used in I4 and I16, while the 8 × 8
DCT is used in I8 mode to further improve the
coding efficiency. In previous H.264/AVC
designs, intra prediction for the baseline and
main profile are well developed for lower specifications like D1 (720 × 480 pixels) and HD
720p. However, there are two main design challenges that lower the efficiency of previous
designs. The first issue comes from the data
dependence between each subblock. According
to the definition of I4 and I8 modes in
H.264/AVC standard, each subblock should be
processed in zig-zag scan order. Since the predictor pixels in the intra prediction are generated from neighboring blocks and are not available
until the neighboring blocks are reconstructed,
each subblock should be processed sequentially.
This data dependence also causes the other
design challenge of low hardware utilization. As
Fig. 5a shows, sequential processing scheduling
makes it difficult to increase the parallelism.
Thus, it costs about 1300 cycles to finish intra
prediction of one MB in a D1-size video under
single-view encoding. However, as mentioned
before, the cycle count available for one MB is
only around 350 cycles under the target HD
multiview specifications.
In order to improve the throughput, the
hybrid open-close loop intra prediction scheme is
proposed to break the data dependence described
above [10]. It is illustrated in Fig. 5b. For subblock boundaries, the original pixels instead of
the reconstructed pixels are used as the intra predictor, and this is the open-loop part. This modi-
fication is based on the assumption that the difference between the original and reconstructed
pixels is very small if the target peak signal-tonoise ratio (PSNR) is higher than 35 dB. In our
target HD multiview environment, this assumption works well. For MB boundaries, the reconstructed pixels are still used as predictors since
these pixels are already reconstructed in the previous MB pipeline stages. The proposed processing schedule is shown in Fig. 5c. Intra prediction
on Blk0 and Blk1 in Fig. 5c can start simultaneously because Blk1 does not need the reconstructed pixels from Blk0. Therefore, the
parallelism of intra prediction can be largely
improved to meet the target HD MVC specifications with little quality loss. However, this openloop scheme cannot be adopted to the
reconstruction step because the original pixels
are not available in the decoder side, and mismatch between the encoder and decoder would
break standard compliance. For this reason, the
reconstruction step is split as a standalone stage.
MBs are reconstructed in a closed-loop manner
in the reconstruction stage. Because of the splitting of the intra prediction and reconstruction
stages, only one best matched mode is required
for reconstruction. Consequently, the cycle budget is enough even under the closed-loop scheme.
FRAME-PARALLEL PIPELINE-DOUBLED
DUAL CABAC
Entropy coding compresses data based on the
probability distribution of symbols, and it plays
an important role in video coding. In the base-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
83
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Data1
Data2
ctx2
ctx1
State (2-symbol)
Symbol
Index Context
Context
modeling Bypass
Side
information
Binary
arithmetic
coding
Range
Range
Low
Low
Bitstream
Output
F
Ctx state
Update
Ctx
comparators
Binarization
BEMaGS
Read state
2*(206:1 table)
CABAC
Syntax
element
A
Output
(a)
Update
Write state
2*(3:1 table)
Ctx state
Output1 Output2
(b)
Task
CABAC
view1
MB2
CABAC
view1
MB1
CABAC
view2
MB1
CABAC
view1
MB3
CABAC
view2
MB2
...
CABAC
view2
MB3
...
Cycles
350
Cycle budget of
other MB pipelines
(c)
(d)
Figure 6. Issues and solutions on the CABAC stage: a) the system overview of CABAC; b) proposed two-symbol arithmetic coder; c) the
frame-parallel scheme of the CABAC stage further improves the symbol rate; d) chip photo of the proposed MVC encoder.
line profile, H.264/AVC adopts context-based
adaptive variable length coding (CAVLC) as the
entropy coder. In the main profile or other
advanced profiles, including the MVC, CABAC
is adopted. CABAC achieves 9 to 14 percent bit
rate savings over CAVLC, but its computation is
much more complicated. Furthermore, due to
the sequential nature of arithmetic coding, the
hardware design makes it extremely difficult to
exploit pipelining or parallel techniques.
Figure 6a shows the block diagram of
CABAC. The inputs of CABAC are syntax elements (SEs) and side information. Syntax elements are the essential data to be coded, such as
MB type, prediction mode, and residues. Side
information, usually the information of neighboring coded blocks, helps to estimate the probability of symbols. These SEs must be
transformed into binary symbols before binary
arithmetic encoding. The adaptive effect is
achieved through the context (ctx) assigned to
the symbol. These ctxs are modeled according to
the SE type, side information, and binary index.
Symbols with the same ctx have similar statistical
properties and use the same adaptive probability
state for estimation. Besides normal arithmetic
coding, bypass mode is introduced to speed up
the encoding process. The symbol along with its
associated ctx and bypass flag enters the binary
84
Communications
IEEE
arithmetic coder. Finally, the arithmetic coder
generates an output bitstream.
Due to the limited cycle budget in the MB
pipeline architecture, an EC engine with a onesymbol arithmetic encoder can only process
about 350 symbols in one MB pipeline stage. As
discussed earlier, this throughput ability is way
below the target HD MVC spec. Therefore, the
multisymbol CABAC architecture is proposed
[11]. The arithmetic coder is duplicated as in
Fig. 6b. For range stage, low stage, and output
stage, two one-symbol PEs are directly cascaded.
However, we cannot simply cascade two onesymbol state stages because they are possibly the
same. The two-symbol state stage is shown on
the right of Fig. 6b. The proposed two-symbol
arithmetic coder may not provide exactly doubled throughput since the throughput depends
on the ctx types. Based on our simulation, the
actual throughput of the proposed two-symbol
coder is 1.94 times larger than the conventional
one-symbol/clock cycle architecture.
Applying the two-symbol CABAC architecture can double the throughput. However, for
some textured MBs, the two-symbol CABAC
architecture still does not meet the throughput
requirement. Based on the analysis from previous work, the critical path increases with the
number of concurrently processed symbols in the
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
arithmetic coder. For our target operating frequency, 250–300 MHz, architectures processing
more than two symbols in parallel are not feasible. Therefore, frame-parallel pipeline-doubled
dual (FPPDD) CABAC is proposed to utilize
frame-level parallelism. Dual CABAC computation cores are adopted, and each CABAC core
has a doubled pipeline cycle budget of 700 cycles.
These CABAC computation cores process in an
interleaved manner as shown in Fig. 6c. Thus,
the frame-parallel scheduling scheme can be
adopted to avoid data dependence between the
two cores. With circuit- and frame-level optimization, the throughput of the proposed
FPPDD CABAC is 3.88 times that of the onesymbol design.
CHIP IMPLEMENTATION
Besides the above algorithm and architecture
optimization, all the other modules, including
the view parallel MB interleaved (VPMBI)
scheduling controller, fractional motion/disparity
estimation [12], and motion/disparity compensation, are also optimized. After adopting all the
proposed solutions, a prototype MVC single
chip encoder was fabricated by Taiwan Semiconductor Manufacturing Company (TSMC) with
90 nm 1P9M process [13]. The chip photo is
shown in Fig. 6d. The core size of the chip is
11.46 mm2 (3.95 mm × 2.90 mm), which contains
1732 kgates. This chip supports both H.264/AVC
Multivew High Profile and High Profile at level
5.1. For multiview video coding, the proposed
MVC chip can support from the full HD 1080p
three views to the HDTV 720p seven views.
According to this view scalability, the processing
ability can be as high as 4096 × 2160p if the view
number is only one. Thus, the proposed chip can
support not only the HD MVC encoder, but also
the quad full HD (QFHD) H.264/AVC singleview encoding.
CONCLUSION
In this article several issues in video encoder
design for 3DTV applications are discussed.
First, the video coding standard development
from 2D to 3D video is introduced. Among these
standards, MVC, an extension profile in
H.264/AVC, provides the best coding efficiency
with a dramatically huge computation requirement. Therefore, very large-scale integrated
(VLSI) hardware acceleration is required to
enable real-time applications. Moreover, the system analysis shows that the previous design
methods used in single video coding have dramatic hardware resource requirements and cannot be employed directly. In order to deal with
these design challenges, solutions for each module in the MVC encoder, including cache-based
and predictor-centered IMDE, hybrid open-close
loop intra prediction, and FPPDD CABAC, are
proposed. After adopting all the proposed algorithm and architecture optimizations, an MVC
single chip encoder is implemented under the
TSMC 90 nm process. By the proposed MVC
encoder design, the target HD MVC specifications can be supported with different view scalability from the 1920 × 1080p full HD three views
to 1280 × 720 HDTV seven views. Furthermore,
the single view QFHD H.264/AVC encoding is
also supported. With the proposed VLSI techniques, real-time 3D video applications become
feasible, and we believe more and more 3D
video consumer products can be realized in the
near future.
IEEE
BEMaGS
F
The frame-parallel
scheduling scheme
can be adopted to
avoid data
REFERENCES
dependency
[1] ISO/IEC MPEG Video and Requirements Group, “Applications and Requirements on 3D Video Coding,”
ISO/IEC JTC1/SC29/WG11 N10857, 2009.
[2] Joint Video Team of ISO/IEC MPEG and ITU-T VCEG,
“Joint Draft 7.0 on Multiview Video Coding,” ISO/IEC
JTC1/SC29/WG11 and ITU-T SG16 Q.6 JVT-AA209, Apr.
2008.
[3] Joint Video Team of ISO/IEC MPEG and ITU-T VCEG,
“WD 1 Reference Software for MVC,” ISO/IEC
JTC1/SC29/WG11 and ITU-T SG16 Q.6 JVT-AA212, Apr.
2008.
[4] Y.-W. Huang et al., “A 1.3TOPS H.264/AVC Single-chip
Encoder for HDTV Applications,” IEEE ISSCC Dig. Tech.
Papers, 2005.
[5] Y. K. Lin et al., “A 242mw 10mm 2 1080p H.264/AVC
High-profile Encoder Chip,” IEEE ISSCC Dig. Tech.
Papers, 2008.
[6] Z. Liu et al., “A Real-Time 1.41w H.264/AVC Encoder
SOC for HDTV 1080p,” IEEE Int’l. Symp. VLSI Circuits
Dig. Tech. Papers, 2007.
[7] C.-Y. Chen et al., “Level C+ Data Reuse Scheme for
Motion Estimation with Corresponding Coding Orders,”
IEEE Trans. Circuits Sys. Video Tech., vol. 16, no. 4, Apr.
2006, pp. 553–58.
[8] L.-F. Ding et al., “Content-aware Prediction Algorithm
with Inter-View Mode Decision for Multiview Video
Coding,” IEEE Trans. Multimedia, vol. 10, no. 8, Dec.
2008, pp. 1553–64.
[9] P.-K. Tsung et al., “Cache-Based Integer Motion/Disparity Estimation for Quad-HD H.264/AVC and HD Multiview Video Coding,” Proc. IEEE Int’l. Conf. Acoustics,
Speech, Signal Process., 2009, pp. 2013–16.
[10] T.-D. Chuang et al., “Algorithm and Architecture
Design for Intra Prediction in H.264/AVC High Profile,”
Proc. Picture Coding Symp., 2007.
[11] Y.-J. Chen, C.-H. Tsai, and L.-G. Chen, “Architecture
Design of Area Efficient SRAM-Based Multi-Symbol
Arithmetic Encoder in H.264/AVC,” Proc. IEEE Symp.
Circuits Sys., 2006, pp. 2621–24.
[12] P.-K. Tsung et al., “Single-Iteration Full-Search Fractional Motion Estimation for Quad Full HD H.264/AVC
Encoding,” Proc. IEEE Int’l. Conf. Multimedia Expo.,
2009, pp. 9–12.
[13] L.-F. Ding et al., “A 212MPixels/s 4096×2160p Multiview Video Encoder Chip for 3D/Quad HDTV Applications,” IEEE ISSCC Dig. Tech. Papers, 2009, pp. 154–55.
between the two
cores. With the
circuit-level and
frame-level
optimization, the
throughput of the
proposed FPPDD
CABAC is 3.88 times
of the one-symbol
design.
BIOGRAPHIES
PEI-KUEI TSUNG (iceworm@video.ee.ntu.edu.tw)
_________________ received his
B.S. degree in electrical engineering and M.S. degree in
electronics engineering from National Taiwan University,
Taipei, Taiwan, in 2006 and 2008, respectively, where he is
working toward his Ph.D. degree in electronics engineering. His major research interests include stereo and multiview video coding, motion estimation algorithms, view
synthesis algorithms, and associated VLSI architectures.
LI-FU DING received his B.S. degree in electrical engineering,
and M.S. and Ph.D. degrees in electronics engineering from
National Taiwan University in 2003, 2005, and 2008,
respectively. In 2009 he joined Taiwan Semiconductor
Manufacturing Company as a principal engineer. His major
research interests include stereo and multiview video coding, motion estimation algorithms, and associated VLSI
architectures.
WEI-YIN CHEN received his B.S. degree in electrical engineering and M.S. degree in electronics engineering from
National Taiwan University in 2005 and 2008, respectively.
In 2007 he was with MIT as a visiting graduate student.
His major research interests include super high definition
and multiview video coding, associated VLSI architectures,
high-level synthesis, and computer architecture.
T ZU -D ER C HUANG received his B.S.E.E. degree from the
Department of Electrical Engineering, National Taiwan Uni-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
85
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
versity in 2005. Now he is working toward his Ph.D. degree
in the Graduate Institute of Electronics Engineering,
National Taiwan University. His major research interests
include the algorithm and related VLSI architectures of
H.264/AVC, and scalable video coding.
YU-HAN CHEN received his B.S. degree from the Department
of Electrical Engineering, National Taiwan University in
2003. He is currently pursuing his Ph.D. degree at the
Graduate Institute of Electronics Engineering, National Taiwan University. His research interests include image/video
signal processing, motion estimation, algorithm and architecture design of H.264 video coders, and low-power and
power-aware video coding systems.
P AI -H ENG H SIAO received his B.S.E.E. degree from the
Department of Electrical Engineering, National TsinhHua University, Hsinchu, Taiwan, in 2007. Now he is
working toward his Master’s degree in the Graduate
Institute of Electronics Engineering, National Taiwan
University. His major research interests include the algorithm and architectures of video coding and neural signal processing.
SHAO-YI CHIEN [S‘99, M‘04] received B.S. and Ph.D.
degrees from the Department of Electrical Engineering,
National Taiwan University in 1999 and 2003, respectively. During 2003 to 2004 he was a research staff member
at Quanta Research Institute, Tao Yuan County, Taiwan.
In 2004 he joined the Graduate Institute of Electronics
Engineering and Department of Electrical Engineering,
National Taiwan University, as an assistant professor.
Since 2008 he has been an associate professor. His
research interests include video segmentation algorithms,
intelligent video coding technology, perceptual coding
technology, image processing for digital still cameras
and display devices, computer graphics, and the associated VLSI and processor architectures. He has published
more than 120 papers in these areas. He serves as an
Associate Editor for IEEE Transactions on Circuits and
Systems for Video Technology and Springer Circuits, Systems, and Signal Processing, and served as a Guest Editor for Springer Journal of Signal Processing Systems in
2008. He also serves on the Technical Program Committees of several conferences, including ISCAS, A-SSCC, and
VLSI-DAT.
86
Communications
IEEE
A
BEMaGS
F
L IANG -G EE C HEN [S‘84, M‘86, SM‘94, F‘01] received B.S.,
M.S., and Ph.D. degrees in electrical engineering from
National Cheng Kung University, Taiwan, in 1979, 1981,
and 1986, respectively. He was an instructor (1981–1986)
and an associate professor (1986–1988) in the Department
of Electrical Engineering, National Cheng Kung University.
In the military service during 1987 and 1988, he was an
associate professor in the Institute of Resource Management, Defense Management College. In 1988 he joined the
Department of Electrical Engineering, National Taiwan University. During 1993 to 1994 he was a visiting consultant
at the DSP Research Department, AT&T Bell Labs, Murray
Hill, New Jersey. In 1997 he was a visiting scholar of the
Department of Electrical Engineering, University of Washington, Seattle. Currently, he is a professor at National Taiwan University. Since 2004 he has also been the executive
vice president and general director of the Electronics
Research and Service Organization (ERSO) in the Industrial
Technology Research Institute (ITRI). His current research
interests are DSP architecture design, video processor
design, and video coding systems. He is a member of the
honor society Phi Tau Phi. He was the general chairman of
the 7th VLSI Design CAD Symposium. He is also the general chairman of the 1999 IEEE Workshop on Signal Processing Systems: Design and Implementation. He has served as
an Associate Editor of IEEE Transactions on Circuits and
Systems for Video Technology since June 1996 and as an
Associate Editor of IEEE Transactions on VLSI Systems since
January 1999. He has been an Associate Editor of the Journal of Circuits, Systems, and Signal Processing since 1999.
He served as a Guest Editor of the Journal of VLSI Signal
Processing Systems for Signal, Image, and Video Technology in November 2001. He is also an Associate Editor of
IEEE Transactions on Circuits and Systems II: Analog and
Digital Signal Processing. Since 2002 he has also been an
Associate Editor of Proceedings of the IEEE. He received
the Best Paper Award from the R.O.C. Computer Society in
1990 and 1994. From 1991 to 1999 he received Long-Term
(Acer) Paper Awards annually. In 1992, he received the
Best Paper Award of the 1992 Asia-Pacific Conference on
Circuits and Systems in VLSI design track. In 1993 he
received the Annual Paper Award of the Chinese Engineer
Society. In 1996 he received the Outstanding Research
Award from NSC and the Dragon Excellence Award from
Acer. He was elected as the IEEE Circuits and Systems Distinguished Lecturer in 2001–2002.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
____________
_______
_____________
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
INTEGRATED CIRCUITS FOR COMMUNICATIONS
An Embedded 65 nm CMOS Baseband
IQ 48 MHz–1 GHz Dual Tuner for
DOCSIS 3.0
Francesco Gatta, Ray Gomez, Young Shin, Takayuki Hayashi, Hanli Zou, James Y.C. Chang, Leonard
Dauphinee, Jianhong Xiao, Dave S.-H. Chang, Tai-Hong Chih, Massimo Brandolini, Dongsoo Koh, Bryan
J.-J. Hung, Tao Wu, Mattia Introini, Giuseppe Cusmai, Ertan Zencir, Frank Singor, Hans Eberhart, Loke
Tan, Bruce Currivan, Lin He, Peter Cangiane, and Pieter Vorenkamp, Broadcom Corporation
ABSTRACT
An embedded CMOS digital dual tuner for
DOCSIS 3.0 and set-top box applications is presented. The dual tuner down-converts a total of
ten 6 MHz Annex B channels or eight 8 MHz
Annex A channels, for a maximum data rate of
320 Mb/s in Annex B and 400 Mb/s in Annex A
mode. The dual tuner exceeds all the stringent
SCTE 40 specifications over the 48–1004 MHz
bandwidth, without using any external components or SAW filters. Enabling technologies are
a harmonic rejection front-end, a low-noise highfrequency resolution PLL, and digital image
rejection. To our knowledge this is the first
reported multichannel broadband tuner embedded in a DOCSIS 3.0 System on a chip implemented in 65 nm pure digital CMOS technology.
MOTIVATION
The convergence of data, audio, and video over
the Internet medium is resulting in ever increasing data bandwidth needs. Due to customer
demand and competition from other services,
cable television (CATV) service providers are
motivated to deliver data rates that far exceed
the capabilities of presently available cable networks. Data Over Cable Service Interface Specification (DOCSIS) is the standard that regulates
high-speed data transfer over cable TV networks. In order to compete with the higher
bandwidth offered, for example, by gigabit passiver optical network (GPON) and very-highrate digital subscriber line (VDSL) services,
cable providers have introduced the DOCSIS 3.0
standard [1]. DOCSIS 3.0 provides bandwidth
increase as well as additional customer flexibility
by bonding together multiple downstream (DS)
data or video channels.
To increase the penetration of digital cable
services in cost-sensitive emerging markets, more
cost-effective and robust tuner integrated circuit
(IC) solutions are necessary. CATV tuners must
operate with good performance over a wide
88
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
range of cable plant conditions, ranging from
high-power-loading scenarios requiring exceptional linearity, to extremely-low-power conditions where linearity must be traded off for
optimized noise performance. Although terrestrial analog transmissions are being replaced
with digital channels in many countries, legacy
analog channels still coexist with digital channels
on the CATV medium. Because the analog channels are usually broadcast at significantly higher
power than the digital channels, the tuner
dynamic range requirements are substantial.
Besides robust operation under different signal
loading scenarios, we desire an architecture that
is easily adaptable to multichannel cable modem
DOCSIS 3.0 applications and can be used for
multituner digital video recorder (DVR) or personal video recorder (PVR) set-top box (STB)
applications.
A cost-effective multichannel multituner solution can be achieved by integrating the tuner,
digital demodulator, MPEG decoder, memory,
and processor core on a single system on chip
(SoC), with considerable savings in power and
system complexity.
This article reports a 65 nm complementary
metal oxide semiconductor (CMOS) dual tuner
embedded in a low-cost low-power DOCSIS 3.0
cable modem SoC that can demodulate up to 8
DS channels with frequency flexibility. In addition, although in the majority of cable plants the
highest spectrum frequency is 864 MHz, cable
providers are extending the CATV bandwidth to
1 GHz to deliver more services. Therefore, we
designed the tuner to operate between 48 MHz
and 1 GHz to allow deployment in all present
and future extended cable plants.
The next section describes the tuner requirements, including the SCTE 40 requirements, in
more detail. We then describe the challenges
related to the tuner design for DOCSIS 3.0,
while the following section describes the dual
tuner architecture. Circuit details are then presented, and the final section reports experimental results.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
DOCSIS 3.0 FREQUENCY
ALLOCATION AND PERFORMANCE
REQUIREMENTS
The CATV radio frequency (RF) input frequency
ranges between 54 and 864 MHz. The DOCSIS
standard calls for 64-quadrature amplitude modulation (QAM) or 256-QAM modulated digital
channels, occupying the same bandwidth as conventional analog TV channels. DOCSIS symbol
rates are 5.36 MHz (corresponding to a 6 MHz
RF bandwidth) for Annex B (United States) and
6.952 MHz (corresponding to an 8 MHz RF bandwidth) for Annex A (worldwide). 256-QAM modulation gives an effective data rate up to 38 Mb/s
in DOCSIS 2.0. DOCSIS 3.0 allows a minimum
data rate of 152 Mb/s by bonding any four channels in a 64 MHz contiguous RF bandwidth. Solutions that allow even more DS channel bonding
and provide more flexibility in the allocated frequency spectrum are preferred by cable operators.
For cable modem applications, the tuner must
meet the DOCSIS specifications [1]. For STB
applications, the requirements for the U.S. market are summarized in the SCTE 40 standard [2].
Most customers currently require SCTE 40 compliance for all CATV applications; therefore, in
this work the tuner design has focused on meeting the more stringent SCTE 40 specifications.
SCTE 40 specifies a set of demodulation scenarios in terms of desired signal power levels,
analog adjacent channel interference (ACI) levels, digital ACI levels, and cable plant loading
with analog and digital channels. The standard
specifies channel impairments, such as incoming
phase noise, echoes, power line hum, and input
signal-to-noise ratio (SNR), which represents the
noise generated by the CATV distribution system.
In a worst-case 256-QAM scenario, the tuner and
demodulator must be able to operate error-free
with a –12 dBmV desired channel with +16 dBc
analog ACI or +12 dBc digital ACI combined
analog and digital loading of the CATV spectrum, 33 dB input SNR, echoes, hum, and added
phase noise. The demodulator requires about 29
dB SNR to operate error-free with margin, which
means that the tuner itself must handle the above
conditions with at least 32 dB SNR. The 64-QAM
low-power worst-case scenario is less stringent
than 256-QAM because error-free video requires
an SNR greater than 23 dB.
TUNER DESIGN CHALLENGES AND
SPECIFICATIONS
A typical U.S. cable TV spectrum, shown in Fig.
1, includes legacy analog channels coexisting
with digital channels, with the analog channels
usually between 54–550 MHz and the digital
channels extending to 864 MHz. This broadband
signal spectrum imposes stringent requirements
on tuner design parameters such as harmonic
rejection, dynamic range, spurious pickup, and
local oscillator (LO) purity. Harmonic rejection
means that the tuner must cope with signals
located at the harmonics of the LO, since they
can be down-converted in the mixing operation
to the same intermediate frequency (IF) as the
Figure 1. Cable spectrum.
desired channel, degrading SNR. Second-order
and third-order distortion products (composite
second order [CSO] and composite triple beat
[CTB]) from many different frequencies can fall
on the desired signal and must be minimized by
using highly linear RF blocks. However, the linearity cannot be obtained at the expense of
noise figure (NF), leading to high dynamic range
requirements. In a broadband RF receiver, particular care must be taken to avoid spurious
noise pickup, in both the signal path and the
synthesizer. This task is especially challenging in
an SoC environment, where the digital noise
sources are too numerous to avoid spurious
noise pick-up by proper frequency planning.
To meet the stringent CATV requirements,
previously reported high-performance silicon
tuner designs have been RF standalone ICs isolated from the noisy SoC environment [3–5]. The
design in [5] uses a dual-conversion architecture,
while [3, 4] use a single-conversion low-IF architecture, with integrated active filters [4] or a system in package (SiP) solution where
high-selectivity filters are implemented in the
package substrate [3]. The first approach uses
expensive external surface acoustic wave (SAW)
filters, and the long receive path results in high
power consumption and overall system complexity. The low-IF SiP architecture reduces power
consumption significantly, due to the external
tracking filters that dramatically reduce the tuner
dynamic range requirements. However, this
approach is not suitable for multichannel applications because of the difficulty in maintaining good
image rejection over many channels. Most important, both architectures are expensive for integration in SoCs, given the use of bipolar CMOS
(BiCMOS) technology [5] and the prohibitive cost
of the SoC package for SiP components [3].
The tuner design is even more challenging for
a DOCSIS 3.0 application due to the channel
bonding requirement. Multiple tuners can be
used to implement a DOCSIS 3.0 system, with
each tuner down-converting a single channel.
This provides full frequency flexibility at the
expense of system cost and power. On the other
hand, a single tuner can down-convert the full 64
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
89
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
32 MHz
Image rejection is
-2 -1
enhanced digitally,
0 +1 +2
Tuner 1 LO
taking advantage of
Tuner 1
the tuner integration
into the SoC,
LPF
2.5-20MHz
LPF
HRM
eliminating the need
Tuner 1 I path
Tuner 1
ADC
for part-to-part
TIA
calibration. An
HR
innovative PLL
PGA VGA
0,-5dB,
-10dB,-15dB
DS AGC
-2
-2
-1
-1
0
0
+1
+1
+2
+2
LO generation 1
architecture achieves
HR
low phase noise and
Tuner 1 Q path
low reference spurs
while providing very
RF PGA
CAM-DS
IMC
RF
fine frequency
resolution.
RF splitter
Tuner 2 Q path
RSSI
LO generation 2
HR
Tuner 2 I path
HR
HR RFFE
Tuner 2
Tuner 2
32 MHz
-2 -1
0 +1 +2
Tuner 2 LO
Figure 2. Dual tuner top-level block diagram.
MHz RF bandwidth with reduced system power
and complexity, but no frequency flexibility since
all the channels need to be contiguous.
TUNER ARCHITECTURE
This article presents a CMOS dual digital CATV
tuner for embedded SoC multichannel applications that simultaneously provides frequency flexibility, low system cost and power, and superior
tuner performance [6]. Two tuners, each able to
select an agile 32 MHz frequency band, can
down-convert ten 6 MHz Annex B channels, or
eight 8 MHz Annex A channels. Figure 2 shows
the top-level block diagram. The RF signal is
amplified by an external low noise amplifier
(LNA), which drives an internal splitter, followed
by the two baseband IQ tuners. Each tuner downconverts five 6 MHz Annex B channels to IF frequencies centered at DC with respect to the
tuner local oscillator (channel 0), +6 MHz (channel +1), +12 MHz (channel +2), –6 MHz (channel –1), and –12 MHz (channel –2). For Annex
A, one tuner down-converts four 8 MHz channels to IF frequencies centered at +4 MHz
(channel +1), +12 MHz (channel +2), –4 MHz
(channel –1), and –12 MHz (channel –2). Channels +1 and +2 are located at the images of
channels –1 and –2, respectively. The baseband
IQ architecture acts as direct conversion for
channel 0 and low-IF conversion for all the other
channels. Any or all channels can be selected for
90
Communications
IEEE
demodulation by the SoC, up to a maximum of
eight. Following the tuner, an eight-channel multirate 64–256–1024-QAM downstream (DS)
receiver separates and demodulates the eight
channels individually. This architecture does not
require a SAW filter, allowing complete integration in the SoC. Each tuner core included in this
RF front-end is designed to meet both DOCSIS
and SCTE 40 standards. To suppress signals at
the odd harmonics of the LO, a harmonic rejection mixer (HRM) is used in combination with a
highly linear RF-tracking filter. Image rejection
is enhanced digitally, taking advantage of the
tuner integration into the SoC, eliminating the
need for part-to-part calibration. An innovative
phase locked loop (PLL) architecture achieves
low phase noise and low reference spurs while
providing very fine frequency resolution.
The dual tuner core has been integrated in a
65 nm CMOS technology DOCSIS 3.0 cable
modem SoC that contains more than 32 million
gates. Embedding a noise-sensitive broadband
tuner in the SoC is particularly challenging
because the double data rate (DDR) clock/data
and million instructions per second (MIPS)
clock fall in the middle of the RF band. A significant amount of noise pickup occurs in the package leads and bond wires. By moving the LNA
off chip, the signal level is raised at the noisy
boundary, and pickup suppression requirements
are relaxed. The external variable gain LNA
(VGLNA) provides up to 20 dB power gain with
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
HR filter
LO1
Flo
3Flo 5Flo
IF OUT
2
7Flo 9Flo
LO2
Flo
3Flo 5Flo 7Flo 9Flo
1
1
IF OUT
1/3
1/5
-1
Flo
LO1
1/7 1/9
1/7 1/9
LO2
2
Flo
LO3
Flo3Flo 5Flo 7Flo 9Flo
1
LO3
7Flo 9Flo
LO=LO1 + 2 LO2 + LO3
(a)
(b)
TIA
CMFB
+
-
LOP_0
LON_0
RFP
W/L
LOP_0
LOP_45
RFN
RFP
IF OUT
LOP_90
LOP_45
LON_45
RFN
2W/L
RFP
LOP_90
LON_90
W/L
RFN
(c)
Figure 3. Harmonic rejection mixer: a) harmonic folding; b) harmonic rejection concept; c) harmonic rejection mixer.
5 dB NF and a stable and high output thirdorder intercept point (OIP3) over a 30 dB gain
range thanks to an internal received signal
strength indicator (RSSI) [7].
TUNER AGC
In a cable environment the tuner input power is
more tightly controlled than in a terrestrial environment where few strong ACI are present. In
cable applications the ACI level is kept low, +16
dBc for 256-QAM, but the number of channels is
very high. The high number of channels and
high-order QAM modulation call for a continuous automatic gain control (AGC) approach. For
this reason our work uses a mixed programmable
Gain Attenuator (PGA)-Variable gain amplifier
(VGA) approach with 50 dB VGA range and 40
dB PGA range. Two separate continuous AGC
loops are used, one at RF in the VGLNA and
one in the tuner baseband in the IF VGA. The
AGC of the VGLNA is controlled by a wideband
internal RSSI, and guarantees a constant power
at the SoC tuner input. The desired channel
power variations are compensated in the IF
VGA, which is driven by the average power measured in the digital demodulator after the ADC.
The RF AGC is optimized for the best noise figure in the different loading scenarios, while the
IF AGC maximizes the ADC loading.
TUNER RECEIVE PATH
The block diagram of the tuner receiver slice
plus splitter and RF front-end (RFFE) is shown
in Fig. 2. A resistor ladder acts as impedance
matching and tapped RF attenuator that optimizes the dynamic range for the two tuners.
Each tuner uses an IQ baseband architecture
with no external filters or SAW filters to minimize external components. This brings design
challenges that must be addressed, as we discuss
in the following subsections.
HARMONIC REJECTION FRONT-END
In a conventional Gilbert cell mixer, the LO
switches are hard driven to improve linearity and
optimize noise figure. However, this generates
LO harmonics, where channels present at the
LO harmonics are folded on top of the desired
channel, degrading SNR, as shown in Fig. 3a. By
using more LO phases and summing them with
the right weighting, it is possible to approximate
a sinusoidal switching that does not contain harmonics other than the fundamental [8]. This
design uses the 0°, 45°, and 90° LO phases to
suppress the third and fifth harmonics (Fig. 3b).
The harmonic suppression is limited by LO
phase inaccuracies and gain mismatch, but careful design and the high-speed capabilities provid-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
91
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
environment the
terrestrial
environment where
few strong ACI are
present. In cable
+1
-2 -1
tuner input power is
controlled than in a
BEMaGS
F
Channel
power
In a cable
more tightly
A
Channel
power
0
Frequency
DC
-2
+1
-1
0
+2
ADC
I
I(n)
LO_I
Flo
Imbalance
canceller
Frequency
ADC
applications the ACI
level is kept very low,
+2
Q(n)
QAM
receiver
Q
LO_Q
+16 dBc for
256-QAM, but the
+
I
number of channels
-
Σ
I*(1-Wa)
Σ
I(n)
-
is very high.
Wa
Q
Wq
+
+
Σ
Q*(1+Wa)
Wa
Σ
-
Q(n)
I 2 - Q2
Wq
I*Q
Figure 4. Single-tap image rejection.
ed by our 65 nm CMOS process guarantee a
minimum harmonic rejection (HR) of 40 dB. In
order to reject the seventh and ninth harmonics
and guarantee a total rejection greater than 65
dB for all the LO harmonics, the embedded
tuner incorporates an RF tracking filter in the
DOCSIS 3.0 SoC. Two source followers isolate
the two tuners, reducing the cross-LO leakage,
and provide a low-ohmic drive of the tracking
filter. If the average frequency of the desired
channels is below 330 MHz, the low-pass tracking filter is selected; otherwise, the mixer is
directly connected to the splitter. This filter is
exposed to the full RF input power, thus requiring extremely high linearity and low noise figure.
To reduce the filter input power, a wide bandwidth first-order pole precedes the filter and
implements a programmable gain optimizing the
dynamic range for the specific spectrum loading
condition that is encountered. A third-order
Butterworth filter is implemented with a complex pole from a biquad and a real passive pole.
An active RC approach is preferred over a GmC approach for its higher linearity and ease of
tuning. The noise figure constraint determines
the resistor value, from which the capacitor size
can be derived and hence the power dissipation.
To achieve the required linearity, the op-amp
unity gain bandwidth needs to be higher than 3
GHz. Such high unity gain bandwidth can be
obtained using the high-speed devices available
in our 65 nm CMOS process. The filter bandwidth ranges between 50 and 330 MHz, and is
tuned with a bank of switched capacitors. Tuning
with capacitors allows the filter to keep the same
92
Communications
IEEE
noise density and hence the same NF for all programmed cutoff frequencies.
HARMONIC REJECTION MIXER
The harmonic rejection mixer (HRM) schematic
is described in Fig. 3c. Three pseudo differential
pairs with resistive degeneration amplify the RF
signal with good NF and linearity. The three
transconductance stages are weighted 1:32:1 and
their outputs are multiplied in a Gilbert multiplier cell by three phases of the LO (at 0°, 45°, and
90°). The current outputs of each mixer are
summed together and injected into the virtual
ground of a trans-impedance amplifier (TIA)
that limits the voltage excursion on the mixer
output, thereby increasing linearity. A first-order
pole in the TIA feedback reduces the dynamic
range for the tuner back-end by attenuating
higher frequency adjacent channels, while the
capacitor area is minimized by means of the virtual ground operation. The TIA gain and first
order pole are programmable to cover singlechannel and multichannel operation, and
increase the overall tuner dynamic range.
TUNER BASEBAND
After the TIA, a fifth-order Butterworth lowpass filter (LPF) serves as an anti-aliasing filter
and attenuates interferers that would saturate
the ADC. The LPF bandwidth is programmable
from 2.5 to 22 MHz to handle single-channel
and multichannel applications. In the multichannel case, the filter must operate with five inband channels whose combined power can be up
to 22 dB higher than the minimum desired chan-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
nel, demanding very high in-band linearity to
minimize cross-channel intermodulation and
clipping. To optimize the trade-off between linearity and NF, the LPF is implemented as two
biquads plus a real pole in the TIA. The biquad
architecture offers more flexibility in the filter
design because the high-Q and low-Q biquads
can be positioned based on the best trade-off
between linearity and noise.
Following the baseband LPF, a PGA and
VGA set the ADC input power, avoiding saturation of the ADC. A 20 dB continuous IF VGA
tracks cable plant power variation and compensates for process, temperature, and voltage
changes by keeping a constant input power at
the ADC input. It maintains full ADC loading
based on the multichannel signal, independent
of the individual channel power values. The
main challenge in this architecture is to keep the
output linearity high under all loading scenarios,
while not degrading the SNR for each individual
channel. Note that low-power desired channels
can coexist with high-power wanted and unwanted channels. Each tuner uses a dual 11-bit 175
Msamples/s pipeline 9.5 ENOB ADC to digitize
the five-channel block.
IMAGE REJECTION
In traditional dual- or single-conversion low-IF
architectures, image rejection (IR) is implemented in the analog domain using external SAW filters or by means of calibration of the analog
imperfections during power-up [3, 5]. In our
design IR is enhanced in the downstream
demodulator, while continuously receiving real
stream data without need for power-up calibration.
In order to demodulate each of the five channels present in the 32 MHz down-converted
band, the corresponding images need to be cancelled. The down-conversion of a block of five 6
MHz channels with the imbalance canceller is
shown in Fig. 4. Due to imperfections of the IR,
image channel information will leak into the
desired channel. For example, channel +1 would
leak on top of channel –1 and vice versa, degrading SNR. In order not to degrade the overall system performance in the presence of all the
channel impairments, the SNR due to the image
must be higher than 45 dB for a 256-QAM modulated signal. Given that the maximum specified
ACI from SCTE 40 is 16 dB for 256-QAM, the
IR must be higher than 61 dB. The finite IR is
mainly due to gain imbalances between the I and
Q path and phase errors in the LO quadrature
generation. As a result of the IQ gain imbalance,
the amplitudes of the I and Q outputs are different, while the effect of an LO quadrature error
is a non-zero product between I and Q. Following the baseband analog-to-digital conversion
(ADC), two correction circuits cancel any gain
and phase imbalance by driving to zero the gain
error (I2 – Q2) and the phase error (I * Q). This
removes any cross-correlation between the two
quadrature paths. In the gain imbalance canceller, the average difference between I2 and Q2
(Wa) is used to equalize the I and Q paths by
multiplying the I-path by (1 – Wa) and the Qpath by (1 + Wa). To eliminate phase imbalances, the product between I and Q (Wq) is
8 bit DAC
DDFS
XTAL
Reference PLL
X 20
Mixer PLL
X 24
1000MHz
2-4GHz
IEEE
F
/4,/8,
/16,/32,/64
LO_Q
Tuner 2
Figure 5. PLL block diagram.
multiplied by the I-path and subtracted from the
Q-path; similarly, Wq is multiplied by Q-path
and subtracted from I-path.
PLL ARCHITECTURE
Demodulation of 256-QAM signals requires an
in-band integrated phase noise better than 50
dBc. A low in-band phase noise can be achieved
by increasing the LO PLL bandwidth. However,
QAM demodulators also require frequency resolution in the tens of kilohertz, which demands a
very low PLL bandwidth. This is the typical
trade-off between in-band phase noise, reference
PLL spurs, and the PLL frequency step size present in classical integer-N PLLs. Fractional-N
PLLs have been commonly used in cable tuners
to widen the PLL bandwidth and obtain fine frequency resolution [3, 5, 9], but they have the
potential to produce fractional spurs. In a dual
conversion architecture fractional spurs can be
tuned out of the desired channel by moving the
up-conversion LO and the down-conversion LO
conveniently. This is not possible in a single- or
direct-conversion architecture such as ours.
Spurs due to digital noise pickup by the PLL are
also an issue. It is sometimes possible to avoid
these spurs by careful frequency planning, but
this is very difficult in a complex SoC, where
there are many asynchronous clock domains.
The PLL for a CATV tuner must have very
low out-of-band (OOB) spurs to avoid folding of
the composite analog and digital loading. In this
article we present a PLL architecture that combines high frequency resolution with a wide loop
bandwidth, without introducing fractional reference spurs (Fig. 5). A 50 MHz low-cost differential crystal oscillator clock is multiplied to
approximately 1 GHz by a fixed frequency reference PLL. This serves as the reference for the
32-bit direct digital frequency synthesis (DDFS)
and an 8-bit digital-to-analog converter (DAC).
The DDFS and DAC combination produces the
reference clock for an integer-N PLL that drives
the down-conversion mixer. This architecture
enables sub-kilohertz LO frequency resolution
while providing a high-frequency reference,
allowing the PLL bandwidth to be increased. A
larger loop bandwidth reduces phase noise,
improves power supply rejection ratio (PSRR),
and shrinks loop filter size, thereby eliminating
the need for any external components. Given
that the minimum reference frequency is higher
than 90 MHz, and the typical PLL bandwidth is
around 500 kHz, reference spurs better than –75
IEEE Communications Magazine • April 2010
Communications
BEMaGS
LO_I
90-180MHz
DAC
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
93
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
55
HR RFPE
Tuner 2
BEMaGS
F
Pin=-12dBmV
Pin=-6dBmV
Pin=+15dBmV
50
Slicer SNR (dB)
Reference PLL
Tuner 1
A
45
40
35
Error free target (29dB)
30
25
0
(a)
200
400
600
Frequency (MHz)
800
1000
(b)
-50
-55
-60
2
-65
ADJ=+16dBc
-70
-75
+4dBmv
SNR=48dB
ADJ=+10dBc
-2dBmv
SNR=41.5dB
-2dBmv
SNR=41.5dB
-80
18A
-12dBmv
SNR=34dB
-85
-12dBmv
SNR=34.5dB
-90
-95
-100
CH -2
CH -1
Center 600 MHz
CH 0
CH +1
3.2 MHz
CH +2
Span 32 MHz
(c)
(d)
Figure 6. Dual tuner die photo and measured results.
dBc are achieved. Measurements indicate similar
performance for all OOB spurs. An active antialias filter following the DAC attenuates unwanted spurs and the DAC images. The mixer PLL
uses an LC-VCO that spans a frequency range
between 2 and 4 GHz. The VCO output is divided by 4/8/16/32 or 64 to cover the RF input spectrum, also generating the quadrature and the 0°,
45°, and 90° phases used in the HR mixers. To
suppress phase modulation spurs in the VCO,
the gain of the VCO is kept low. The 2 –4 GHz
VCO range is implemented using a dual tank
LC and an array of switched capacitors. The 1
GHz reference PLL uses a low-noise ring-based
VCO with 6 MHz bandwidth.
EXPERIMENTAL RESULTS
A micrograph of the dual tuner is shown in Fig.
6a. In Fig. 6b the measured 256-QAM slicer SNR
is shown for three input power levels: –12 dBmV
(minimum power for SCTE 40), –6 dBmV (midpower), and +15 dBmV (high power). The minimum power SNR indicates the tuner NF and the
margin from SCTE 40 error-free target. The midpower SNR indicates the tuner dynamic range.
The high-power SNR represents the tuner peak
SNR. The peak SNR drops at higher frequencies
due to phase noise limitations. The single-channel
94
Communications
IEEE
SNR tests have been performed with Internet traffic on. Since there is no frequency where the SNR
is substantially degraded due to the presence of
spurs, we can conclude that the tuner immunity to
signal path and PLL spurs is very high.
A DOCSIS 3.0 sample operation is described
in Fig. 6c with five channels received simultaneously. The power for channel 0 and channel +2
is –12 dBmV; for channels +1 and –1 it is –2
dBmV (+10 dBc ACI); while for channel –2 it is
+4 dBmV (+16 dBc ACI). Channels +2 and –2
are images of each other, and they must be separated before demodulation. Therefore, the SNR
of channels 0 and +2 is a good measure of the
IR achieved. By noting that there is hardly a difference in SNR between channel 0 and channel
+2, we can conclude that over-all IR performance is exceeding 60 dB. 1024 QAM modulation can provide a 25 percent throughput
improvement over 256 QAM modulation, further increasing the DOCSIS 3.0 data rate at the
expense of more tuner dynamic range. In Fig. 6d
a constellation for an 847 MHz, +15 dBmV
input power clean channel with 47 dB SNR is
shown. Phase noise for the 1 GHz LO is –99
dBc/Hz at 10 kHz offset, while it is better than
–122 dBc/Hz at 1 MHz offset. The in-band integrated phase noise at 1 GHz is 0.2° root mean
square (rms) from 5 kHz to 10 MHz. PLL out-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
Gupta
ISSCC ’07 [9]
Tourret
JSCC ’07 [3]
Stevenson
ISSCC ’07 [5]
This work
48–864 MHz
48–864 MHz
48–864 MHz
48–1000 MHz
have different tuner
5–7 dB
requirements. In the
Cable modems for
VoIP applications
Input range
NF
4–7 dB
5 dB
6–8 dB
case of an electrical
256 QAM sensitivity
PN @ 10 KHz
–22 dBmV
–90 dBc
–90 dBc
–85 dBc
–100 dBc
power outage, the
cable modem must
be able to operate in
CTB/CSO
–53/–115 dBc
–57/–57 dBc
–65/–60 dBc
–66/ –64 dBc
IR
61 dB
62 dB
75 dB
>62 dB
Max input power
44 dBmV
50 dBmV
External filter
NO
YES (SiP)
YES
NO
SoC integration
NO
NO
NO
YES
Multiple channels
NO
NO
—
YES
DS channels
8
Power (DOCSIS-VoIP)
540 mW, 1.8 V
—
—
800 mW, 2.5 V
Tuner power (SCTE 40)
—
780 mW, 3.3 V
1500 mW, 5 V
1100 mW, 2.5 V
LNA = 400 mW
RX = 400 mW
PLL = 250 mW
REFPLL = 50 mW
ADC = 200 mW
Process
0.18 +m CMOS
0.25 +m SiGe (SiP)
0.35 +m SiGe
65 nm CMOS
Area (one tuner)
25 mm2
5.7 mm2 (die)
9 mm * 9 mm (SiP)
7.25 mm2
5 mm2
battery backup
mode and guarantee
continuity of phone
service. Power
consumption for the
cable modem SoC
and the tuner must
be minimized at the
expense of
performance.
Table 1. Performance summary and comparison with published literature.
of-band (OOB) and reference spurs are better
than –75 dBc over the full cable band. With 135
+15 dBmV (–32 dBm) analog channels, the
measured CTB, CSO and cross modulation
(XMOD) are better than –66 dBc, –64 dBc, and
–60 dBc respectively.
The tuner can operate in two power modes:
SCTE 40 mode when used in set-top box applications and DOCSIS mode when it is used in a
cable modem application. The SCTE 40 low
power worst-case condition for –12 dBmV input
power, +16 dBc ACI, 33 dB cable SNR, and
cable plant loading is met with 1.5 dB worst case
minimum margin. In a fully loaded test like
SCTE40, PLL spurs can fold all the high power
ACI on top of the desired channel. Good margin
over the specifications testify for the purity of
the tuner LO.
Cable modems for VoIP applications have
different tuner requirements. In the case of an
electrical power outage, the cable modem must
be able to operate in battery backup mode and
guarantee continuity of phone service [10].
Power consumption for the cable modem SoC
and the tuner must be minimized at the expense
of performance: typical ACI levels are lower
than 10 dBc, cable input SNR is 34 dB, and
cable loading is lower than SCTE 40. With
reduced power consumption, the tuner meets
DOCSIS requirements over frequency with similar margins as SCTE40 operation. Finally, a performance summary with a comparison with the
state of the art is shown in Table 1.
CONCLUSIONS
An embedded CMOS digital dual tuner for
DOCSIS 3.0 and set-top box applications has
been presented. The dual tuner can down-convert a total of ten 6 MHz Annex B channels or
eight 8 MHz Annex A channels, for a maximum
data rate of 320 Mb/s in Annex B and 400 Mb/s
in Annex A mode. The dual tuner exceeds all
the stringent SCTE 40 specifications over the
48–1004 MHz bandwidth, without using any
external components or SAW filters. Enabling
technologies are a harmonic rejection front-end,
a low-noise high-frequency-resolution PLL, and
digital image rejection.
ACKNOWLEDGMENTS
The authors would like to thank J. Brannon, S.
Freville, R. Nguyen, R. Whitehead, and D.
McMullin for their support.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
95
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
REFERENCES
[1] CableLabs, “Data Over Cable Service Interface Specifications, DOCSIS 3.0, Physical Layer Specification,” CM-SPPHYc3.0-I07-080522, May 22, 2008.
[2] SCTE Eng. Committee, “Digital Cable Network Interface
Standard,” ANSI/SCTE 40, 2004.
[3] J. R. Tourret et al., “SIP Tuner with Integrated LC Tracking Filter for Both Cable and Terrestrial TV Reception,”
IEEE J. Solid-State Circuits, vol. 42, no. 12, Dec. 2007,
pp. 2809–21.
[4] J. van Sinderen et al., “A 48 MHz to 860 MHz Digital
Cable Tuner IC with Integrated RF and IF Selectivity,”
IEEE ISSCC Dig. Tech. Papers, Feb. 2003, pp. 444–45.
[5] J. M. Stevenson et al., “A Multi-Standard Analog and
Digital TV Tuner for Cable and Terrestrial Applications,”
IEEE ISSCC Dig. Tech. Papers, Feb. 2007, pp. 210–11.
[6] F. Gatta et al., “An Embedded 65 nm CMOS Baseband
IQ 48 MHz–1 GHz Dual Tuner for DOCSIS 3.0,” IEEE J.
Solid-State Circuits, vol. 44, no. 12, Dec. 2009, pp.
3511–25.
[7] D. Manstretta et al., “A Highly Linear Broadband Variable Gain LNA for TV Applications,” IEEE CICC ‘07,
Sept. 2007, pp. 531–34.
[8] J. Weldon et al., “A 1.75 GHz Highly Integrated Narrow-Band CMOS Transmitter with Harmonic Rejection
Mixers,” IEEE J. Solid-State Circuits, vol. 36, Dec. 2001,
pp. 2003–15.
[9] M. Gupta et al., “A 48-to-860 MHz CMOS Direct Conversion TV Tuner,” IEEE ISSCC Digest Tech. Papers, Feb.
2007, pp. 206–7.
[10] CableLabs, “PacketCable 1.5 Specifications MIBs
Framework Specification,” PKT-SP-MIBS1.5-I02-070412,
Apr. 12, 2007.
ADDITIONAL READING
[1] ITU-T G.984.1, “Series G: Transmission Systems and
Media, Digital Systems and Networks, Gigabit-Capable
Passive Optical Networks (GPON): General Characteristics,” Mar. 2008.
[2] ITU-T G. 993.2, “Series G: Transmission Systems and
Media, Digital Systems and Networks, Very High Speed
Digital Subscriber Line Transceivers 2 (VDSL2),” Feb.
2006.
[3] I. Vassiliou et al., “A 65 nm CMOS Multistandard,
Multiband TV Tuner for Mobile and Multimedia Applications,” IEEE J. Solid-State Circuits, vol. 43, no. 7, July
2008, pp. 1522–33.
BIOGRAPHIES
FRANCESCO GATTA (fgatta@broadcom.com)
____________ received his Laurea
and Ph.D. degrees in electrical engineering from the University of Pavia, Italy, in 1998 and 2001, respectively. His Ph.D.
research focused on CMOS highly integrated receivers for
UMTS applications and low-power CMOS LNA. From 2001 to
2002 he was with Valence Semiconductor working on CMOS
GPS products. Since 2002 he has been with Broadcom Corporation, Irvine, California, in the Analog-RF Group where he
has been working on PLL for SERDES applications, CTF for
hard disk drives, and tuners for cable and satellite applications. Currently he is leading the development of all the
CATV and DTV CMOS integrated tuners. His main interests
are in RF CMOS design, tuner architectures, and broadband
systems. He is the named inventor of three U.S. patents.
RAY GOMEZ received his B.S degree in biomedical engineering
from Case Western Reserve University in 1981, his M.S.E.E
degree in electrical engineering from Stanford University in
1982, and his Ph.D. degree from the University of California,
Los Angeles (UCLA) in 1993. He worked at TRW, Inc., Redondo Beach, California, from 1982 to 1986, where he designed
high-performance frequency synthesizers. His research at
UCLA, from 1986 to 1993, involved analog CMOS integrated
circuits for disk-drive read channels. From 1993 to 1995 he
was a member of the disk drive read channel group at Cirrus
Logic, Austin, Texas. He joined Broadcom Corp. in 1995,
where he has focused on CMOS RF circuits for cable and
broadcast television tuners, DBS satellite set-top box tuners,
and, more recently, MoCA home multimedia networking. He
was named a Broadcom Fellow in 2006 for his contributions
to tuner design, and has over 30 issued patents.
YOUNG J. SHIN received his B.S. degree in electrical engineering and computer science from the University of California
(UC) Berkeley, in 1994 and his M.S. degree in electrical
96
Communications
IEEE
A
BEMaGS
F
engineering from UCLA in 1996. Since 1996 he has been
with Broadcom Corp., mostly designing CMOS integrated
tuners for DBS satellite, QAM cable, and fiber optic (MoCA)
applications. He is currently a manager of engineering at
Broadcom focusing on research and development of current and future generation tuners for QAM and MoCA settop box and modem SoCs.
T AKAYUKI H AYASHI received B.A.Sc. and M.A.Sc. degrees in
electrical engineering from the University of Toronto,
Ontario, Canada, in 1995 and 1997, respectively. From
1997 to 2000 he was with ATI Technologies, Thornhill,
Ontario, working on the development of ADC for video
applications. In 2000 he joined Broadcom Corp., where he
is currently a principal design engineer. His research interests are in analog and mixed-signal circuit design for communication systems.
HANLI ZOU received his B.S. degree in electronic engineering
from Tsinghua University, Beijing, China, in 1996, and his
M.S. and Ph.D. degrees in electrical engineering from UCLA
in 2000 and 2003, respectively, with a focus on integrated
circuits and systems for wireless broadband communication systems. From 2001 to 2003 he worked as senior ASIC
designer at Innovics Wireless Corporation, Los Angeles, California, designing baseband chipsets for a diversity enabled
WCDMA transceiver. Since 2003 he has been with the
Broadband VLSI Group, Broadcom Corp. as a principal scientist, working on advanced transceiver chip design for
cable, satellite, home networking, and terrestrial digital TV
and set-up box. His research interest includes system
design and VLSI implementation for high-speed communication, with an emphasis on digital compensation for
RF/analog front-end impairments, synchronization, channel
estimation, equalization, and diversity processing.
J AMES Y.C. C HANG received his B.Sc. degree (magna cum
laude) from UC Irvine, and his M.Sc. and Ph.D. degrees from
UCLA in 1990, 1992, and 1998, respectively, all in electrical
engineering. He was a co-recipient of the Best Paper Award
at the 1995 ESSCIRC, and the Jack Raper Award for Outstanding Technology Directions at ISSCC ’97. He received the
Outstanding Ph.D. Student Award from UCLA for the year
1997–1998. He is currently with Broadcom Corp. developing
cable and satellite set-top-box SoC embedded tuners. He is a
named inventor on seven issued patents.
LEONARD DAUPHINEE earned a B.Eng. in electrical engineering
from Dalhousie University, Halifax, Canada, in 1987. From
1987 until 1989 he designed underwater data telemetry
products at Vemco Ultrasonics. He returned to Dalhousie
University to earn an M.A.Sc. in electrical engineering in
1991, specializing in real-time DSP architectures. From 1991
until 1993 he was the engineering manager of RF communications design at the Applied Microelectronics Institute.
He began doctoral studies at Carleton University, Ottawa,
Canada, with a focus on RFIC design in 1993. In 1998 he
joined Broadcom Corporation as a staff scientist working on
broadband tuner products and earned his Ph.D. from Carleton University in 2003. He is presently the senior engineering manager of RF tuner development at Broadcom and is
named as inventor on 29 issued or filed U.S. patents.
J IANHONG X IAO [S‘04] received his B.Sc. degree from the
Department of Computer Science and Technology, Peking
University, China, in 2001, and his Ph.D. degree from Texas
A&M University in 2007. He worked at Analog Devices as
design engineer intern in summer and fall 2004 and summer 2005. He joined Broadcom as a staff scientist in 2007.
His main research interests cover analog and RF front-end
design for broadband communication systems.
TAI-HONG CHIH received his diploma in electrical engineering
from National Tsing-Hua University in 2001, and his M.S degree
in electrical and computer engineering from Carnegie Mellon
University in 2005. He has worked in the RF tuner department
at Broadcom since 2005. His interests include RF/analog circuits
and tuner/broadband communication systems.
DAVE (SUNG-HSIEN) CHANG received B.S. and M.S. degrees in
communication engineering from National Chiao-Tung University, Hsin Chu, Taiwan, R.O.C., in 1992 and 1994,
respectively. He received his Ph.D. degree in electrical engineering from the UCLA, where he was working on
microwave and millimeter-wave circuits in 2000. In 2000
he joined Broadcom Corp., where he is involved in designing integrated circuits for CMOS standalone and SoC
tuners for cable, set-top box, and satellite applications.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
MASSIMO BRANDOLINI received his Laurea (summa cum laude)
and Ph.D. degrees in electrical engineering and computer
science from the University of Pavia in 2002 and 2006,
respectively. During his Ph.D., he worked on analog and RF
ICs for wireless communications in CMOS and BiCMOS
technologies, with focus on RF front-ends for cellular applications and multistandard radios. In 2003 he was with
Agere Systems, Allentown, Pennsylvania, as an internship
student, working on the design of a fully integrated CMOS
FM transmitter. He joined Broadcom Corp. in January
2006, where he is a senior staff scientist working on highly
integrated tuners for cable, satellite and MoCA systems. He
is a co-recipient of the IEEE Journal of Solid-State Circuits
2003 Best Paper Award and the IEEE Symposium on VLSI
Circuits 2005 Best Student Paper Award.
JUO-JUNG HUNG received B.S. and M.S. degrees in electrical
engineering from National Taiwan University, Taipei,
R.O.C., and his Ph.D. degree in electrical engineering from
the University of Michigan, Ann Arbor, where he was
involved with the development of SiGe RFIC and RF microelectromechanical systems (MEMS) for microwave and millimeter-wave applications. He is currently a senior staff
design scientist with Broadcom Corp., where he has
focused on CMOS analog and RF circuits for satellite, cable,
and terrestrial television receivers.
DONGSOO KOH [S‘89, M‘98] received B.S. and M.S. degrees
in electronics engineering from Sogang University, Seoul,
Korea, in 1989 and 1991, respectively, and a Ph.D. degree
in electrical engineering from UCLA in 1997. From 1997 to
1999 he was with BethelTronix, Inc, Cerritos, California,
where he developed RF ICs for cordless phones and GPS
systems. From 1999 to 2006 he was with Skyworks Solutions, Inc. (formerly Conexant Systems, Inc.), Irvine, California, where he worked on RF IC designs for various mobile
standards including CDMA 2000 and GSM/EDGE. Since
2006 he has been with Broadcom Corp. His work has
focused on RF tuner development for broadband communication products.
Tao Wu received his B.S. degree in mechanical engineering
from Harbin Institute of Technology, China, in 1992, and his
M.S. degree in electrical engineering from the University of
Arizona in 1999. Currently, he is with Broadcom’s RF tuner
group and involved in RF and mixed signal circuit designs for
broadband applications. From 2005 to 2006 he worked for
Skyworks. From 2000 to 2005 he worked at
Motorola/Freesclae, Libertyville, Illinois, where he was involved
in RF and mixed signal designs for wireless applications.
Mattia Introini received his Laurea degree in electrical engineering and computer science from the University of Pavia
in February 2006. He joined STMicroelectronics in 2006 as
an analog IC designer; during this period he worked on
analog and RF ICs for hard disk drive applications in CMOS
technology, with focus on the read channel RF front-end.
In October 2006 he joined Broadcom Corp. as a design
engineer. He has been working on highly integrated tuners
for cable, satellite, and MoCA systems.
G I U S E P P E C U S M A I [S‘05] received his Laurea and Ph.D.
degrees in electrical engineering and computer science
from the University of Pavia in 2003 and 2007, respectively. His Ph.D. research was focused on CMOS and bipolar
TX/RX front-ends for WPAN communications (UWB). In
2005 he was with National Semiconductor, Santa Clara,
California, as an internship student, working on a CMOS
low-power highly integrated transmitter for Zigbee. Since
January 2007 he has been with Broadcom Corp., where he
develops highly integrated CMOS tuners for cable and
satellite systems.
ERTAN ZENCIR received B.Sc. and M.Sc. degrees in electrical
and electronics engineering from Middle East Technical
University, Ankara, Turkey, in 1995 and 1997, respectively,
and a Ph.D. degree in electrical engineering from Syracuse
University, New York, in 2003. Between 2004 and 2005 he
worked as an assistant professor in the Department of
Electrical Engineering, University of Wisconsin-Milwaukee.
From 2005 to 2006 he was with the RFIC design team of
Nokia Mobile Phones, San Diego, California, as a senior
RFIC design engineer, where he specialized in SiGe BiCMOS
cellular RF transceiver IC design for CDMA2000. Since 2006
he has been with the RF Tuner Development Department
of Broadcom Corp., where he is currently a senior staff
design scientist developing analog and RF CMOS ICs for
IEEE
BEMaGS
F
cable, satellite receivers, and MoCA transceivers. His interests include analog and RF IC design for broadband and
wireless communications.
F RANK W. S INGOR received his B.A.Sc. degree in electrical
engineering from the University of Western Ontario, Canada, in 1991 and his M.A.Sc. degree from the University of
Toronto, Ontario, Canada in 1994. In 1994 he joined the
Data Converter Group at Maxim Integrated Products, Sunnyvale, California. In 1999 he joined Broadcom Corp.,
working in the Mixed Signal Group, and presently is managing Broadcom RF Tuner Development, Austin, Texas.
HANS EBERHART received B.S. and M.S. degress in electrical
engineering from the University of Pennsylvania, Philadelphia, in 2000 and 2001, respectively. Between 2002 and
2004 he was working toward his Ph.D. in electrical engineering at UCLA. In 2004 he joined Broadcom Corp., where
he is involved in analog and mixed-signal circuit design.
LOKE K. TAN received his B.S. degree from the University of
Houston, Texas, in 1987, and his M.S. and Ph.D. degrees
from UCLA in 1992 and 1995,respectively, all in electrical
engineering. He was a consultant to PairGain Technologies
in 1992. Since 1993 he has been with Broadcom Corp.,
where he works in the Communications Systems and IC
Design Group. He has been involved in the design of QAM
transceivers for HDTV applications. His interests include
digital signal processing, digital communications, and highperformance CMOS circuit design. He received the 1995
Best Paper Award from the IEEE Journal of Solid-State Circuits for the paper entitled “A 200 MHz Quadrature Digital
Synthesizer/Mixer in 0.8 mm CMOS.”
L IN H E received her Ph.D. degree from the University of
Pennsylvania in 2005. From 2005 to 2007 she was with
Sarnoff Corporation, Princeton, New Jersey. She joined
Broadcom Corp. in 2007 as a senior staff scientist. She is
currently involved with system design for broadband communication systems.
BRUCE CURRIVAN is technical director, Broadband Communications at Broadcom Corp., where he develops advanced
modem architecture. He holds a B.S. from Cornell University and an M.S.E. from Princeton University, both in electrical engineering and information sciences. Prior to
Broadcom he worked at RCA Astro-Electronics Division,
Stanford Telecommunications, Inc., and Wavespan Corporation. He served as Chairman of the IEEE 802.14 Cable
Modem Physical Layer Subgroup, and was a major contributor to the DOCSIS specifications. He is the author of many
designs, papers, book chapters, and patents in the areas of
modem design and adaptive equalization.
PETER CANGIANE received Bachelor’s and Master’s of Science
degrees in electrical engineering from Polytechnic Institute
of New York in 1986 and UCLA in 1988, respectively. He
was with TRW Corporation from 1986 until 2000, where
he specialized in multirate digital signal processing. Since
2000 he has been with Broadcom Corp.’s Broadband Cable
Business Unit. He is the named inventor of one U.S. patent.
PIETER VORENKAMP received his M.S. degree in electrical engineering from Twente University, Enschede, The Netherlands. From 1989 to 1996 he was with Philips Research
Laboratories, Eindhoven, The Netherlands, researching
high-speed data converters in BiCMOS processes. From
1996 to 1997 he was with Philips Semiconductors, Caen,
France, responsible for the development of high-speed
data acquisition systems for video and instrumentation
applications. In 1997 he joined Broadcom Corp., where he
has held progressively senior engineering management
positions. In 1999 he managed the Analog and RF Microelectronics group in Irvine, California, responsible for the
analog part of all mixed-signal chips for application in digital communication systems, ranging from high-fidelity
audio front-ends to RF tuner subsystems for cable and TV
applications. In 2008 he managed Broadcom’s Power Management Business Unit and successfully introduced the first
standalone Power Management and Energy Management
product line for mobile and other battery powered SoCs
into the market. Most recently he is responsible for operations engineering, including foundry engineering, product
engineering, product test, and assembly within Broadcom’s
Operations and Central Engineering organization. He is an
author of many international publications and is a named
author of almost 50 U.S. patents.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
97
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
INTEGRATED CIRCUITS FOR COMMUNICATIONS
Integrated Electronic System Design for
an Implantable Wireless Batteryless
Blood Pressure Sensing Microsystem
Peng Cong, Medtronic Inc.
Wen H. Ko, Case Western Reserve University
Darrin J. Young, University of Utah
ABSTRACT
A wireless, batteryless, less invasive blood pressure sensing microsystem based on an instrumented circular cuff has been developed for advanced
biological research. The proposed sensing technique avoids vessel penetration and substantially
minimizes vessel restriction due to the soft cuff
elasticity. The integrated electronic system design
is presented with emphases on the design trade-off
and system considerations. The measurement
results demonstrate full functionality of the
microsystems with real-time high-fidelity blood
pressure sensing capability under wireless data
telemetry and adaptive RF powering conditions.
INTRODUCTION
Cardiovascular diseases are the number one
cause of death and disability in the United States
and most European countries. Long-term in vivo
blood pressure monitoring is critical for treating
cardiovascular disease and hypertension. Recently, DNA sequencing of small laboratory animals
together with real-time monitoring of blood pressure as well as other vital signals has become a
critical research tool to identify genetic susceptibility to diseases and to potentially develop new
treatment methods [1]. However, the small blood
vessel size of those animals, less than 1 mm in
diameter for major arteries, introduces a significant design challenge for the blood pressure
monitoring microsystem; there is no good solution for its long-term in vivo monitoring to date.
The most common technique used in small
laboratory animal monitoring relies on an invasive
catheter-tip transducer, which requires a complex
surgical procedure and could cause increased
blood pressure, blood clotting, and reduced sensitivity over time. Furthermore, discrete electronics
are typically employed to implement the implant
system, resulting in a large form factor with excessive power dissipation, which in turn calls for a
bulky ferrite-based radio frequency (RF) coil for
external RF powering or battery recharging, thus
limiting the accuracy of the measured biological
signals due to post-implant trauma for long-term
98
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
monitoring. Therefore, a miniature lightweight
long-term blood pressure sensing implantable
microsystem with wireless data telemetry and
adaptive RF powering capability is highly desirable to capture the vital signs of a free roaming
small laboratory animal housed in its home cage,
as depicted in Fig. 1.
A LESS INVASIVE BLOOD PRESSURE
MONITORING SYSTEM
A 3D illustration of a wireless, batteryless, less
invasive implantable blood pressure monitoring
system is illustrated in Fig. 2. The system employs
an instrumented elastic cuff wrapped around a
blood vessel, which is not shown in the picture, to
sense real-time blood pressure waveforms. The
elastic cuff is made of bio-compatible silicone
material and is filled with low-viscosity silicone
oil with an immersed micro-electro-mechanical
system (MEMS) pressure sensor and integrated
electronic system. The MEMS sensor measures
the pressure waveform in the cuff coupled from
the expansion and contraction of the vessel. A
rigid silicone isolation ring is used to decouple
the sensing cuff, which is located at the structural
center, from environmental variations in the animal’s body. Because of the softness of the sensing cuff’s outside wall, the pressure inside the
sensing cuff is susceptible to environmental variations, such as muscle and tissue movement, without the isolation ring. The isolation ring is
designed so that an air cavity between the isolation ring and the sensing cuff’s outside wall is
formed upon completion of the fabrication process. As a result, the sensing cuff’s outside wall
can move freely, responding to the blood vessel
pressure waveform. In the figure an additional
thin metal layer wraps around the isolation ring
for further suppression of environment variations. The overall structure has a radius of 3.2
mm and a width of approximately 4 mm to establish an adequate contact area with a blood vessel
to be inserted in the middle of the cuff through
the opening/closing slot. A surgical suture will be
applied to secure the cuff position after vessel
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
External RF powering
and data acquisition
Data receiving antenna
F
A wireless powering
system with an
adaptive power
control capability is
required to provide a
B.P. waveform
sufficient and stable
energy to power the
Anim
implant electronics in
al ca
RF powering coil
Blood pressure monitoring system
ge
a varying magnetic
field. The RF-DC
RF
power converter also
extracts clock
information from the
RF powering signal.
Figure 1. In vivo real-time wireless batteryless blood pressure monitoring microsystem.
insertion. The radius of the sensing cuff’s inside
wall is designed to be approximately 0.5 mm with
an outside sensing cuff wall radius of approximately 0.9 mm, thus adequate for laboratory animals with an artery diameter of around 1 mm.
The silicone sensing cuff can be fabricated by a
conventional machining and molding process [2].
The measured waveform represents a downscaled version of the vessel blood pressure waveform with a typical scaling factor of 10 percent
and is processed by a nearby low-power integrated
electronics system exhibiting a dimension of
approximately 2 × 2 × 3 mm3, consisting of MEMS
sensor interface circuitry, an analog-to-digital converter (ADC), and a system configuration and
control unit, followed by wireless data telemetry to
an external receiver. Post-implant calibration can
be performed to reconstruct the vessel blood pressure waveform from the measured data. An adaptive RF-DC power converter is incorporated to
provide sufficient and stable energy to the system
implanted in an untethered animal as shown in
the overall system diagram in Fig. 3. The spiral
coil for RF powering is located outside the cuff
with sealed feed-through connections to the electronics. The proposed sensing technique avoids
vessel occlusion, bleeding, and blood clotting associated with the conventional catheter-tip-based
approach. Furthermore, the sensing cuff is made
of soft silicone material. The restrictive effect on
the vessel is therefore substantially minimized,
thus making it suitable for long-term monitoring.
INTEGRATED ELECTRONIC
SYSTEM DESIGN
System integration of integrated electronics and
micro-fabricated sensors can provide a unique
solution to realize an implantable microsystem
with small size, light weight, and reliable sensing
capability. A MEMS capacitive pressure sensor is
designed and incorporated in the sensing cuff to
convert blood pressure information to capacitance
variation. A capacitance-to-voltage (C/V) converter
is thus required in the integrated electronic system
to convert the capacitance variation to a voltage
signal. The output of the C/V converter is then digitized by an ADC for wireless data transmission.
Wireless RF powering and data telemetry are also
incorporated in the microsystem to eliminate the
need for external wire connections and any bulky
battery. The microsystem can therefore be used to
monitor freely moving small laboratory animals to
obtain reliable measurements without stressinduced distortion due to wire connections or large
system size. RF powering has been widely used for
biomedical implants, where both transmitting and
receiving units are properly placed at a fixed distance from each other with a constant RF power
coupling coefficient. However, in this research the
receiving unit is implanted inside a freely moving
laboratory animal, resulting in a continuously
changing RF power coupling. Therefore, a wireless
powering system with an adaptive power control
capability is required to provide a sufficient and
stable energy to power the implant electronics in a
varying magnetic field. The RF-DC power converter also extracts clock information from the RF
powering signal. An on-chip power sensor is
designed to detect the received RF power level
and generates a 1-bit power-level feedback signal.
The 1-bit power-level feedback signal together with
the digitized blood pressure information is then
transmitted to an external receiver by an on-chip
oscillator-based frequency shift keying (FSK) transmitter with an off-chip inductor achieving the function of an antenna. The high-Q off-chip inductor is
employed for low power dissipation. The received
blood pressure information is used for real-time
blood pressure monitoring. The 1-bit power-level
feedback signal is used to control an external
power amplifier supply voltage to realize adaptive
RF powering capability.
To realize a functional system, the electronics
should be properly designed to achieve performance
requirements, such as pressure resolution, system
dynamic range, on-chip DC power level via RF
powering, sensor data transmission rate, and wireless communication range. Miniaturization can be
achieved by integrating various building blocks onto
a single chip as well as keeping individual building
blocks small by employing an optimal architecture.
In addition, each building block should be designed
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
99
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Metal isolation ring
m
5m
3–
m
2
3.
m
Silicone
isolation Coil for RF powering
ring
Sensing cuff
outside wall
MEMS pressure sensor
with Interface Electronics
immersed in silicone oil
Figure 2. 3D configuration of the blood pressure sensing microsystem.
immune to interference for robust performance. A
fully differential architecture is used for the sensor
interface electronics and ADC design for adequate
common-mode signal rejection. A low-power design
is important for implantable systems, where the
power source is highly limited. A low-power design
can be achieved by choosing proper system architectures and circuit implementations. A capacitive pressure sensor, instead of a piezoresistive pressure
sensor, is selected for the overall microsystem design
to minimize power dissipation. Interface electronics
are designed to operate in a weak inversion region
for further power reduction. It is difficult to provide
a high supply voltage for wireless and batteryless
microsystem operation. Therefore, low-voltage
design techniques become essential. For a 1.5 +m
complementary metal oxide semiconductor (CMOS)
process selected for the prototype design with VTP
and VTN of approximately –1 V and 0.7 V, respectively, a 2 V power supply level is chosen as a tradeoff between circuit design complexity and RF
powering constraints. The 2 V power supply will be
generated by an on-chip RF-DC power converter. A
2.5 V supply will also be generated to control the
MOS field effect transistor (MOSFET) without dissipating any DC power. The design trade-off of the
MEMS pressure sensor, C/V converter, cyclic ADC,
and transmitter is presented in detail in this section.
The adaptive RF-DC powering will be covered by
other papers.
MEMS CAPACITIVE PRESSURE SENSOR
The MEMS capacitive pressure sensor was
designed for the prototype microsystem design
due to its miniature size, high sensitivity, low
temperature dependence, zero DC power dissipation, and time stability. The fabricated sensor
exhibits dimensions of 0.4 × 0.5 × 0.4 mm3 with a
measured nominal capacitance value of approximately 2 pF and a sensitivity of approximately
0.8 fF/mmHg [2]. The device size is adequate to
be enclosed in the blood pressure sensing cuff.
CORRELATED DOUBLE SAMPLING C/V
CONVERTER WITH AUTOMATIC OFFSET
CANCELLATION SCHEME
Interface electronics are required to convert sensor
capacitance variation, which represents the blood
pressure information, into analog voltage prior to
100
Communications
IEEE
A
BEMaGS
F
digitization. The C/V converter needs to be
designed for a sensing resolution of 100 aF over a
1 kHz bandwidth and an allowed overall full-range
capacitance variation of 200 fF, which corresponds
to a dynamic range of 11 bits. An automatic offset
cancellation scheme is designed as shown in Fig. 3
to allow a single-ended pressure sensor with a wide
range of nominal capacitance value to be used,
thus greatly simplifying the MEMS fabrication process; it can also effectively suppress the output offset voltage. During the initial phase of the circuit
operation, the digitally controlled reference capacitor array at the amplifier input is cycled through to
find a reference capacitance closely matched to the
sensor nominal capacitance.
1/f noise from input transistors of a front-end
amplifier used in the C/V converter is typically the
dominant noise source, and can effectively be suppressed by correlated double sampling (CDS) or
chopper stabilization techniques [3, 4]. The CDS
technique, which was originally introduced to
reduce the noise produced in charge-coupled
devices, is an effective method to suppress low-frequency noise, DC offset, and charge injection
effects in switched-capacitor circuits. A typical CDS
amplifier samples the amplifier output voltage twice
per clock cycle: first in a reset phase and second in
an evaluation phase. The offset and low-frequency
noise, such as 1/f noise, at these two sampling
instances are nearly constant due to a relatively
high sampling frequency. Therefore, they are highly
correlated, and can be eliminated by performing a
subtraction operation between the two samples.
The sampling frequency needs to be designed much
higher than 1/f noise corner frequency with a typical
ratio of 10. A disadvantage of CDS is that the
uncorrelated thermal noise is doubled due to the
double sampling. The amplifier must be designed
to meet the thermal noise requirements.
Chopper stabilization is another commonly
used technique to reduce offset and low-frequency
noise in precision analog circuit design. First, the
sensor information is modulated to a high frequency by a stimulation clock, where the stimulation frequency needs to be much higher than 1/f
noise corner frequency. The modulated sensor signal is then mixed by a clock exhibiting the same
frequency as the stimulation frequency, followed
by a low-pass filter to obtain the original sensor
signal. In this process the 1/f noise and DC offset
from the charge amplifier are modulated to the
stimulation frequency by the mixer and then eliminated by the low-pass filter. A comparison
between chopper stabilization and CDS techniques shows that CDS excels in a number of
aspects for this application. First, switched-capacitor-based CDS is more compatible with a sampled
data system. Second, instead of being modulated
to a high frequency by the chopper stabilization
technique, the DC offset and 1/f noise are eliminated by the CDS technique at the amplifier output, thus increasing the allowable signal swing
range, which is critical for the low supply voltage
design. Finally, chopper stabilization typically
requires a mixer and a low pass filter, which result
in additional power consumption and silicon area.
There are three noise contributors for the
pre-amplifier circuit. The first contributor is the
switch thermal noise associated with the switches
connected to the left plates of the sensor capaci-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Integrated circuit
Voltage doubler
L2
Adaptive VDD
VHIGH
VLOW
C2
Skin
4 MHz
Class-E
PA
On-ship DC voltage
regulators
CLK
extraction
RF power
input
Power-level sensor
L1
C1
1110001
Quantized 1-bit
power-level
VR
MEMS capacitive
pressure sensor
11- bit
cyclic
ADC
C/V
VDD_Analog_2V
VDD_Digital_2V
2-channel
combiner/
polarity
generator
0101110
Inductor as
transmitter
antenna
φ2
φ1
φRST
CII
MEMS capacitive pressure sensor
VS
φ1
φ2
φRST
VB
ICMFB
CFB
128C
D7
4C
D2
2C
D1
1C
D0
CH
CI
+ –
CFB
+ –
OCMFB
– +
CI
φ2
VCM
OCMFB
– +
CH
CII
To ADC
VCM
φ2
φRST
φ1
φ2
8 bit counter
Digital controlled
reference capacitors
Digital
B.P. and
power-level
feedback
information
FSK
transmitter
Digital B.P.
information
CDS C/V converter with automatic offset cancellation
Cs
2.5V
CalEn
Range detector
2.5-Gain stage
Figure 3. Electrical system design architecture.
tor, C S , and the reference capacitor, C R . As a
common-mode signal, this noise is highly suppressed. The second contributor is the noise
associated with the switches connected between
the pre-amplifier input and output. This noise is
eliminated by the CDS scheme as part of the
pre-amplifier 1/f noise and DC offset. Therefore,
the dominant noise source is the third contributor, the amplifier thermal noise. A fully differential telescopic architecture with p-type MOS
(PMOS) input transistors is chosen as the main
amplifier for its low noise and low power dissipation compared to other architectures, such as a
folded-cascode amplifier architecture. The folded-cascode architecture requires additional biasing transistors, thus resulting in higher current
dissipation and noise contribution. Compared to
a fully differential telescopic architecture, the
folded-cascode topology improves input common-mode range. However, in the proposed C/V
converter, input common-mode feedback
(ICMFB) is employed to set the pre-amplifier
input to a fixed voltage level, thus rendering the
advantage associated with the folded-cascode
architecture. The amplifier is biased in weak
inversion with a 6 +A bias current, achieving an
——
input-referred noise of 23 nV/3Hz, which corresponds to a capacitance sensing resolution of
approximately 75 aF or a pressure resolution of
0.1 mmHg over 1 kHz. With a typical cuff scaling
factor of 10 percent, a vessel blood pressure sensing resolution of 1 mmHg is expected. The weakinversion design technique is used to achieve the
transconductance requirement at a reduced current dissipation with a resulting slow circuit
speed. The reduced circuit speed, however, is
suitable for typical biomedical applications with
signal bandwidth typically ranging from a few
Hertz to a few kilohertz.
An ICMFB circuit is incorporated with the
pre-amplifier design to minimize the input common-mode shift caused by the stimulation clock,
hence minimizing offset due to any mismatch of
parasitic capacitances and potential output drift
over time. To maintain the same input commonmode voltage level with the stimulation voltage
of V S, a common-mode charge of V S(C S + C R)
must be compensated for by the ICMFB circuit.
A CFB of 5 pF is selected to sufficiently compensate for the pre-amplifier input common-mode
shift with a maximum sensor capacitance, CS, up
to 5 pF. The bandwidth of the ICMFB is
designed to be approximately 80 kHz by using a
bias current of 1.2 +A. 80 kHz, which is 10 times
higher than the pre-amplifier closed-loop bandwidth, is chosen to effectively compensate for
the pre-amplifier input common-mode variation.
CYCLIC ADC
There are six commonly used ADC architectures: flash ADC, integrating ADC, successive
approximation (SAR) ADC, pipeline ADC,
sigma-delta ADC, and cyclic ADC. The following outlines the advantages as well as disadvantages of each architecture. Flash ADC is for
ultra-high-speed application with a typical reso-
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
101
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
10
b1
Range
detector
Vref/4
01
b0
-Vref/4
00
φ3
φ2
S/H
C4
φ1
– +
Vin+
Vin-
+–
φ1
C4
φ9
φ2
φ3
φ8
φ9
φ10
φ4
φ6
φ5
φ6
VCM
Vout+
φ11
φ12
C2
C1
+ –
φ4d
VRef
VCM
0
φ7
VoutVCM
φ6
MDAC
C3
– +
C1
C2
φ5
φ6
φ4
C3
φ4d
φ12
φ11
φ10
0
VCM
VRef
φ8
Figure 4. Schematic of an 11-bit cyclic ADC.
lution up to 8 bits. The ADC exhibits high power
consumption and large size, mainly due to 2N – 1
comparators required for an N-bit resolution.
On the contrary, integrating ADC is typically
used for monitoring DC and low-frequency signals with high resolution. However, a slow conversion rate is the drawback for this type of
ADC, thus making it inadequate for the proposed application. SAR ADC is intended for
medium to high resolution (8- to 16-bit resolution) with medium speed, which is suitable for
the proposed biomedical application. However,
this ADC is based on a successive approximation
scheme, thus requiring a precisely matched
capacitor array. It is difficult to achieve 11–12bit accuracy without an extensive layout effort.
In addition, a large capacitor array leads to a
large silicon area. Sigma-delta ADC is processinsensitive to the first order and is efficient in
terms of power consumption as well as silicon
area, but cannot readily be used to process multiplexed input signals due to the lack of direct
correspondence between an analog input sample
and digital output bits without employing a decimation filter on-chip. Pipeline ADC is intended
for medium- to high-speed applications. It has
become the most popular ADC architecture for
sampling rates from a few megasamples per sec-
102
Communications
IEEE
A
BEMaGS
F
ond up to 100 Msamples/s, with resolutions from
8 to 16 bits. This ADC could be used for implementing the proposed microsystem; however, the
area would be large due to the multiple stages
required with this architecture. Cyclic (algorithmic) ADC is well known to achieve low power
consumption and high resolution in a small silicon area [3, 5], which is suitable for the intended
application. This type of converter can be
designed without calling for well matched capacitors. The detailed schematic of an 11- bit cyclic
ADC is illustrated in Fig. 4. The multiplying digital-to-analog converter (MDAC) input voltage
is sampled twice, followed by an exchange of
sampling capacitor and integrating capacitor for
a gain of two. Class A/AB amplifiers are used in
the ADC for low power consumption. The overall ADC consumes a current of 6.3 +A, achieving a signal-to-noise-and-distortion ratio
(SNDR) of 65 dB, which is equivalent to an
effective number of bits (ENOB) of 10.5.
OSCILLATOR-BASED TRANSMITTER
A wireless data telemetry system is a crucial building block in an implant microsystem to minimize
risk of infections and signal interference associated with hard-wire connections. Passive or active
data telemetry could be used for any implantable
mircrosystem. Passive data telemetry utilizes
reflected impedance from a secondary circuit load
seen by a primary circuit for data communication.
By utilizing backscatter modulation to provide a
reflected impedance change at an external coil circuitry, technically, it consumes zero power. However, due to the small internal coil connected to the
microsystem, the coupling coefficient is much
lower than 0.1 percent. In addition, animal movement and title angle result in much weaker coupling with a large variation range. Therefore, this
requires a complex demodulator circuit with high
resolution and sensitivity, which is not practical for
the current system. An alternative choice, active
data telemetry, is chosen for the system, and provides a reliable wireless data link independent of
animal position and title angle. The challenge of
transmitter design for implantable devices is the
stringent power consumption requirement.
FSK modulation is chosen for a reliable wireless data link and system simplicity compared to
amplitude shift keying (ASK) and binary phase
shift keying (BPSK) counterparts. A carrier frequency of 433 MHz, which is in the industrial, scientific, and medical (ISM) radio band, is chosen
for the prototype microsystem testing and characterization due to the availability of commercial
receivers in this band. In addition, 433 MHz is
adequate for the implant application, low enough
to avoid increased transmission loss through live
tissues and high enough for a small coil antenna
to be employed for overall system miniaturization. An inductor and capacitor (LC)-tuned oscillator with cross-coupled configuration is used for
its design simplicity and power efficiency; a
schematic is presented in Fig. 5. This configuration also allows a single inductor to be employed
for the oscillator. The inductor is implemented by
a high-Q off-chip inductor with a size of 1.8 ×
1.24 × 1.24 mm3 to ensure low power consumption with an optimal DC bias current of 120 +A
to develop a steady state differential oscillation
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
2V
M3
M4
V+
V-
MC
C
MC
VTX
M2
M1
Vbias
120 μA
Figure 5. Oscillator-based transmitter for data
telemetry.
amplitude of 0.6 V. The inductor also serves the
function of an antenna for wireless data transmission. PMOS capacitors, MC, are used as variable
capacitors, which are controlled by VTx, presenting Manchester-encoded digital data. The size of
the PMOS is designed so that 250 kHz frequency
separation is realized with a VTx amplitude of 2.5
V, which is adequate for a data rate of 48 kb/s in
the prototype microsystem. The remaining bandwidth can be used for other sensing channels,
which can potentially be integrated with the blood
pressure sensing microsystem.
WIRELESS BATTERYLESS
MICROSYSTEM IN VIVO
CHARACTERIZATION
The integrated electronic system has been fabricated in a 1.5 mm CMOS process, exhibiting a
chip area of 2.2 × 2.2 mm 2, and the fabricated
overall microsystem exhibits a weight of approximately 430 mg including the metal isolation ring,
which was approximately 100 mg [2]. The weight
of 430 mg was less than 0.2 percent of a typical
laboratory rat body weight with a value between
200 g and 400 g. The total system noise under
truly wireless and batteryless conditions is measured to be 750 +V RMS referred to the ADC
input, closely matching the designed value, which
is equivalent to a resolution of 75 aF or 0.1
mmHg over 1 kHz with a dynamic range of 60
dB. The total power consumption is 300 +W. The
microsystem was implanted in a laboratory rat at
the right carotid artery. A commercial cathetertip transducer was inserted into the left carotid
artery as a reference for comparison with an
implant, shown in Fig. 6a. During characterization, the microsystem was powered by an external
coil fabricated on a PCB positioned underneath
the rat with a size of 15 cm × 20 cm driven by a
Class-E power amplifier. The measured digital
blood pressure information was transmitted to a
nearby external commercial receiver by an onchip oscillator-based transmitter through FSK
modulation in the setup shown in Fig. 6a. The
worst case bit error rate (BER) is 10–3, which is
limited by the commercial receiver used for testing. The maximum transmission distance between
the microsystem and the receiver is approximately 15 cm, which is adequate for this application.
The measured blood pressure waveforms by the
wireless batteryless microsystem and the reference catheter-tip transducer are presented in
Figs. 6b and 6c, respectively. The two waveforms
are well matched in shape, exhibiting similar
blood pressure characteristics with a calculated
correlation coefficient of 95 percent. A scaling
factor of 0.15 can be calculated from the measured waveform. The parameter can be used to
reconstruct the blood pressure waveform in the
vessel from the recorded waveform from the
monitoring cuff. A heart rate of approximately
220 beats/min can also be extracted from both
waveforms. A reliable blood pressure waveform
can also be monitored under wireless and batteryless conditions when the rat is freely running
in its own cage after recovery from the surgery.
The data measured for 24 hours after implant
exhibits a noise level increased by approximately 3
dB. The VCO-based transmitter center frequency
was also decreased from 430 to 426 MHz over 24
hours. Further investigations show that the
enhanced noise was likely caused by vapor penetration through the silicone coating to the electrical
connections of the MEMS sensor and integrated
circuit (IC) chip. The frequency drift was also likely caused by vapor penetration to the electrical
connection traces, which is equivalent to adding a
30 fF capacitor to LC tank. The results tell us the
biological environment is typically the limiting factor for a high-performance implantable system. An
improved packaging technique is expected to
ensure the system performance over time.
IEEE
BEMaGS
F
It is expected that
the proposed
sensing technique
with microsystem
engineering design
will be desirable for
future human-based
health monitoring.
Other sensing
channels, such as
temperature, EKG,
activity, can be
readily integrated in
the system.
CONCLUSION AND FUTURE WORK
A wireless and batteryless implantable blood
pressure sensing microsystem is demonstrated
for real-time blood pressure monitoring. The
design and trade-offs of the integrated electronic
system have been presented in detail. The prototype microsystem was successfully implanted in
laboratory rats to measure real-time blood pressure waveforms achieving a resolution of 0.1
mmHg over 1 kHz with a dynamic range of 60
dB. The demonstrated wireless implantable technology will become an important research tool
for system biology research.
It is expected that the proposed sensing technique with microsystem engineering design will
be desirable for future human-based health
monitoring. Other sensing channels, such as
temperature, EKG, and activity, can readily be
integrated in the system to achieve a multichannel monitoring unit, which is an ongoing project
in our research group. Two blood pressure sensing cuffs can also be positioned with a fixed distance along an artery to obtain meaningful
real-time blood flow information for cardiovascular disease study [6].
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
103
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Receiving antenna
Class-E amplifier
driving external coil
F
Right carotid artery
Microsystem for
blood pressure
monitoring
Laboratory rat
Catheter-tip
transducer
External coil
(15 cm × 25 cm)
(a)
800
160
770
760
750
740
10
20
30
Time (s)
40
50
60
0
770
760
750
Pressure (mmHg)
780
6.2 mmHg
Output code
130
120
0
790
740
730
140
10
30
Time (s)
20
40
50
60
160
150
38 mmHg
730
150
38 mmHg
6.2 mmHg
Output code
780
Pressure (mmHg)
790
140
130
120
0
0.1
0.2
0.3
0.4
0.5 0.6
Time (s)
0.7
0.8
0.9
1
0
0.1
0.2
0.3
0.4
0.5 0.6
Time (s)
0.7
0.8
0.9
1
(c)
(b)
Figure 6. Wireless batteryless blood pressure monitoring microsystem laboratory rat evaluation: a) microsystem implant; b) measurement
from the blood pressure microsystem; c) measurement from the catheter-tip transducer.
REFERENCES
[1] B. Hoit et al., “Naturally Occurring Variation in Cardiovascular Traits among Inbred Mouse Strains,”
Genomics, vol. 79, no. 5, May 2002, pp. 679–85.
[2] P. Cong, W. H. Ko, and D. J. Young, “Wireless
Implantable Blood Pressure Sensing Microsystem
Design for Small Laboratory Animals Monitoring,” Sensors Materials, vol. 20, no. 7, 2008, pp. 327–40.
[3] P. Cong, W. H. Ko, and D. J. Young, “Low Noise mWatt
Interface Circuits for Wireless Implantable Real-Time
Digital Blood Pressure Monitoring,” CICC ’08, San Jose,
CA, Sept. 2008, pp. 523–26.
[4] C. C. Enz and G. C. Temes, “Circuit Techniques for
Reducing the Effects of Op-Amp Imperfections: Autozeroing, Correlated Double Sampling, and Chopper Stabilization,” Proc. IEEE, vol. 84, no. 11, Nov. 1996, pp.
1584–1614.
[5] P. Li et al., “A Ratio-Independent Algorithmic Analogto-Digital Conversion Technique,” IEEE J. Solid-State Circuits, vol. 19, no. 6, Dec. 1984, pp. 828–36.
[6] K. Takahata et al., “A Wireless Microsensor for Monitoring Flow and Pressure in a Blood Vessel Utilizing a
Dual-inductor Antenna Stent and Two Pressure Sensors,” 17th IEEE Int’l. Conf. MEMS, 2004, pp. 216–19.
BIOGRAPHIES
P ENG C ONG (pengcon77@gmail.com)
______________ received his Ph.D.
degree from the Department of Electrical Engineering and
Computer Science at Case Western Reserve University
104
Communications
IEEE
(CWRU)in 2009. His research focuses on MEMS sensors,
mixed-signal IC design, and microsystem integration for
biomedical and harsh environmental applications. He is a
senior IC design engineer for Medtronic Inc. Minneapolis,
Minnesota, working on next-generation implantable medical devices.
WEN H. KO [F] received Ph.D. degrees in electrical engineering from Case Institute of Technology, Cleveland, Ohio, in
1956 and 1959 respectively. He has been an assistant,
associate, and full professor of electrical engineering and
biomedical engineering, at CWRU since 1959, 1962, and
1967, respectively. He became a Professor Emeritus in electrical engineering of CWRU in July 1993 and is active in
research on MEMS and biomedical implants including
micropackage and thin film power supplies. He is a fellow
of the American Institute of Medical and Biological Engineering.
DARRIN J. YOUNG received his B.S., M.S., and Ph.D. degrees
from the Department of Electrical Engineering and Computer Sciences at the University of California at Berkeley in
1991, 1993, and 1999, respectively. He joined the Department of Electrical Engineering and Computer Science at
CWRU in 1999 as an assistant professor. In 2009 he joined
the Electrical and Computer Engineering Department at the
University of Utah as a USTAR associate professor. His
research interests include MEMS design, fabrication, and
integrated analog circuits design for wireless sensing,
biomedical implant, communication, and general industrial
applications.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
CALL
SYNCHRONIZATION
OVER
ETHERNET
FOR
AND
A
BEMaGS
F
PAPERS
IP
IN
NEXT-GENERATION NETWORKS
Network synchronization plays a central role in digital telecommunications. It determines the quality of most services provided by the network operator. Traditionally, synchronization has been distributed across telecommunications networks
using the TDM links for which the networks were designed (i.e. E1 and DS1 circuits).
Several fixed and mobile Operators are migrating to a next-generation network (NGN) with IP packet-switched network
infrastructure. Ethernet transport is becoming increasingly commonplace. This trend is driven by prospected lower operative costs and by the convergence between fixed and mobile services. However, migrating trunk lines to IP transport
poses significant technical challenges, especially for circuit emulation and synchronization of network elements.
The evolution of communications networks towards packet-switching has increased interest in the distribution of synchronization using packet methods. This has a twofold impact:
1. Synchronization distribution over packet networks using packet-based methods has become a focus of activity in
thestandards bodies (ITU-T G.8261/2/3 and IEEE 1588);
2. The traditional model, in which synchronization distribution is engineered carefully for optimal performance, may
give way to scenarios in which there is a greater expectation of ad-hoc synchronization quality achieved without
as much need for provisioning as has traditionally been the case — similar to Ethernet "plug and play".
The latter point widens the scope of interest in synchronization beyond specialists to the wider audience of telecommunications engineers. A striking example is the distribution of synchronization to next-generation wireless base-stations,
which are connected to the core network only via packetswitched networks, but still require highly accurate synchronization to meet service quality expectations.
SCOPE OF CONTRIBUTIONS
Authors from industry and academia are invited to submit papers for this special issue of IEEE Communications Magazine
on next generation synchronization, including synchronization distribution over Ethernet and IP networks. The scope of
the issue includes, but is not limited to the following topics of interest:
•
•
•
•
•
•
Key emerging standards in the area of synchronization over packet networks
Timing scenarios in various network applications
Tutorial papers describing packet network characteristics which impact synchronization transport
Implementation challenges in achieving high-quality synchronization using the new technologies
Carrier and vendor experience in deployment
Measurement techniques for characterizing and qualifying packet-based synchronization
The special issue is expected to have 5-7 papers.
SUBMISSION GUIDELINES
Articles should be tutorial in nature, with the intended audience being all members of the communications technology
communities. They should be written in a style comprehensible to readers outside the specialty of the article.
Mathematical equations should not be used (in justified cases up to three simple equations are allowed). Articles should
not exceed 4500 words. Figures and tables should be limited to a combined total of six. The number of references is recommended to not exceed 10 (maximum 15). Complete guidelines for preparation of the manuscript are posted at
http://dl.comsoc.org/livepubs/ci1/info/sub_guidelines.html. Please send a PDF (preferred) or MSWORD formatted paper via
Manuscript Central (http://mc.manuscriptcentral.com:80/commag-ieee). Register or log in, and go to the Author Center.
Follow the instructions there. Select "February 2011/Synchronization over Ethernet and IP in Next-Generation Networks".
Manuscript Deadline:
Notification of acceptance:
Final paper submission:
Publication date:
May 31, 2010
September 30, 2010
November 30, 2010
February 2011
GUEST EDITORS
Stefano Bregni
Politecnico di Milano, Dept. of Electronics and Information
Piazza Leonardo da Vinci 32, 20133 Milano MI, Italy
Email: _______________
bregni@elet.polimi.it
Communications
IEEE
Ravi Subrahmanyan
National Semiconductor Corp.
One Stiles Road, Suite 305, Salem, NH, 03079, USA
Email: ____________________
ravi.subrahmanyan@ieee.org
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
ACCEPTED FROM OPEN CALL
Power Line Communication Networks
for Large-Scale Control and
Automation Systems
Gerd Bumiller, iAd GmbH
Lutz Lampe, University of British Columbia
Halid Hrasnica, Eurescom GmbH
ABSTRACT
Power line communications uses the existing
power line infrastructure for communication
purposes. While the majority of recent contributions have discussed PLC for high-data-rate
applications like Internet access or multimedia
communication serving a relatively small number
of users, in this article we are concerned with
PLC as an enabler for sensing, control, and
automation in large systems comprising tens or
even hundreds of components spread over relatively wide areas. Typical examples of such systems are energy management (Smart Grid) and
facility automation systems. We provide a discussion of the communication network requirements common to such systems and present
transmission concepts for PLC to make use of
the existing infrastructure resources (i.e., power
lines) to meet these requirements.
INTRODUCTION
Already during World War II, power lines were
considered as a means of data transmission [1].
The main usage of power line communications
(PLC) has been by electricity companies for
sending control signals at low rates and in several home automation products. It was only recently, spurred by the deregulation of the
telecommunication and energy market in the
late nineties, that communication over power
lines has received wider attention and is perceived by many as a viable alternative or valuable complement to other wired or wireless
communications systems.
This is particularly true for Internet access
and indoor local area networks (LANs), where
application of so-called broadband PLC is considered. Broadband PLC assumes service provision for multimedia applications consuming
larger data rates and serving a limited number
of users, and its new popularity is evidenced
by two special issues on PLC-based LANs and
access networks in this magazine in 2003 [2,
3], and the recent developments in standard-
106
Communications
IEEE
0163-6804/10/$25.00 © 2010 IEEE
ization for high-speed PLC systems within the
IEEE [4].
The single main advantage of PLC over other
wired communication solutions is the existence
of a power line infrastructure. This, for example,
allows electricity companies to retrofit their
power line networks for communication purposes at little additional cost. In fact, the energy distribution grid is perhaps the most ubiquitous
infrastructure worldwide, and its extremely high
penetration opens the door for a plethora of
applications supported by PLC. Alongside the
aforementioned applications, especially the use
of PLC for advanced energy management has
experienced a strong boost. Examples of this
trend include the recently completed research
and development project Real-Time Energy
Management via Powerlines and Internet (REMPLI), which involved nine partners from five
European countries and was funded by the European Union (see www.rempli.org), and the Powerline Intelligent Metering Evolution (PRIME)
project launched by the Spanish electric utility
Iberdrola and joined by industrial partners from
Europe and the United States, whose aim is the
specification of an open and non-proprietary
PLC-based telecommunications architecture that
“could meet the future requirements on customer real time interfacing and smart grid evolution” (see www.iberdrola.com/suppliers/
SmartMetering for a White Paper). More gener__________
ally, the ubiquity of power distribution lines renders PLC an excellent candidate for industrial
command-and-control and facility automation
systems, in which a common communication network connects a large number of devices that
are spread over a wide area. We collectively
refer to such systems as large-scale control and
automation systems in the following.
The design and performance requirements
for PLC in large-scale control and automation
systems are decidedly different from those for
PLC in access or indoor systems. Different network parameters, such as geographical coverage
or number of network nodes, and different application-related features, such as size of data pack-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
ets or maximal response time, make it necessary
to apply different transmission concepts. In the
first part of this article we describe these requirements and derive the necessary features of an
enabling PLC network. This is followed by a
description of transmission techniques that have
been developed for such PLC networks in the
past few years [5–9]. We focus on the physical
and medium access control (MAC) layer techniques, which are closely linked to the use of
PLC as communication technology. The gist of
our discussion is that conceptually fairly simple
single-frequency networking together with flooding of messages are attractive methods for large
PLC networks. While such an approach has been
advocated in wireless ad hoc networks [10], we
submit that it is also suitable for PLC networks
(with no mobility) to enable service guarantees.
To support this observation, we finally present a
quantitative comparison of flooding and routing
based on performance parameters specifically
relevant to the considered application scenarios.
Most of the presented material originates from
work conducted for energy management systems,
under the umbrella of REMPLI, and airfield
ground lighting automation systems. Therefore,
even though we keep the ensuing description
generic for the most part, we repeatedly refer to
these two specific applications.
CONTROL AND AUTOMATION
SYSTEMS USING PLC
We start by considering the two mentioned
application examples for large-scale control and
automation systems to motivate PLC as a very
attractive solution and concretely show what is
required of a PLC network.
APPLICATION EXAMPLES
Energy Management Systems — The energy
management system of the future foresees the
implantation of considerable intelligence into
the distribution grid, which essentially renders it
a situation-aware network of interconnected sensors and actuators. These intelligent grids of
tomorrow have gained global attention under
the label Smart Grid [11]. The realization of the
Smart Grid concept entails the existence of a
reliable communication network, which likely
integrates several communication carriers. Due
to the inherent availability of power lines as carriers and the resulting advantages with respect to
installation costs, we expect that PLC will play a
prominent role in the implementation of Smart
Grid.
Let us consider a few typical examples of
Smart Grid functionalities to illustrate the
requirements for the communication network.
The first example is real-time pricing to balance
energy consumption and moderate peak loads.
The key components for real-time pricing are
intelligent energy meters and the possibility of
communication with a central data collection
and control station. Assuming that every household is equipped with such a meter, the PLC
network between meters and the common transformer station can easily include 300 nodes or
more. If the PLC network extends beyond the
transformer station and to the medium-voltage
layer, the network size can grow to thousands of
nodes. Fast access to individual meters is also
needed to enable advanced customer service.
For example, in a recent tender a Dutch electric
utility required that their call center can access
90 percent of all meters within five seconds. In
Germany the national regulator Bundesnetzagentur considers the possibility of real-time
switching between electric utilities, and the vendors of billing software work on solutions for
customers to find the presently least expensive
provider. Again, fast access to meters is an
important element for these solutions.
Another Smart Grid functionality is the management of energy distribution using a supervisory control and data acquisition (SCADA)
system. SCADA sensors permanently monitor
the grid load and report to a control center,
from which open/close commands are sent to
switches to adapt the distribution structure to
the dynamics of energy generation and consumption. Such operations become more frequent
with increasing decentralization of energy generation, and they need to be executed reliably and
in real time in order to maintain grid stability.
Thus, SCADA imposes strong reliability and
real-time requirements on PLC.
IEEE
BEMaGS
F
The energy
management system
of the future
foresees the
implantation of
considerable
intelligence into the
distribution grid
which essentially
renders it a situationaware network of
interconnected
sensors and
actuators.
Airfield Ground Lighting Automation Systems — Modern airfield ground lighting (AGL)
automation systems enable individual lamp control and monitoring of sensors deployed at airfields. Such functionalities are needed to meet
the latest recommendations by national and
international regulators to enhance safety of aircraft ground movement and to aid visual guidance systems. Figure 1 illustrates a typical wiring
topology of an AGL system. Devices such as
lamps and microwave sensors are arranged in a
ring structure of typically between 3 and 15 km
connected to a constant current supply via transformers.
PLC is a cost-effective and elegant solution
to enable communication between the airport
tower and the ground lighting system, particularly for existing airport infrastructures where
build-up of new dedicated communication networks would be expensive. The communication
network has to bridge considerable distances and
also connect across power-electric components,
especially transformers, which are not designed
for high-frequency communication signals. In
terms of signal flow, we note that all data communication needs to go through a central node
directly connected to the tower (Fig. 1). Furthermore, the reaction time, the round-trip delay of
a signal between the tower and a lamp, is critical. Since the communication channels are timevariant, the PLC system needs to permanently
monitor the communication quality to all network nodes to guarantee a certain maximal reaction time. The time variance of the channels is
due to different current steps of the regulator,
variable loads in the circuit, crosstalk from other
rings that often run parallel over several kilometers, and even weather conditions. Hence, the
PLC system needs to be sufficiently robust with
respect to channel variations. In addition, a node
failure must not affect communication to other
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
107
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Automation and grid
A
BEMaGS
F
Sensor
management
systems have mainly
a master-slave
Tower
Data link
Remote
node (slave)
Remote
node (slave)
Remote
node (slave)
Remote
node (slave)
Remote
node (slave)
Remote
node (slave)
structure, because
these applications
are strictly
hierarchically
organized. In facility
management,
Constant
current
regulator
PLC
central
node
(master)
client-server
networks are
becoming popular.
Sensor
Figure 1. Illustration of the ring structure of an airfield ground lighting automation system. A typical ring has
a length between 3 and 15 km, and includes between 10 and 200 remote nodes (lamps, sensors).
nodes. Hence, redundant signal paths between
the central and other network nodes are mandatory for the PLC network.
REQUIREMENTS FOR THE PLC INFRASTRUCTURE
The described application examples are representative in that the PLC network needs to connect a large number of devices like switches,
sensors, meters, and lamps that are distributed
over a relatively wide area. In the following we
attempt to categorize the typical requirements
for the PLC infrastructure in order to support
large-scale control and automation systems.
Network Coverage and Data Flow — Application protocols for metering, automation, facility or grid management support point to
multi-point communication with mainly short
data packets. Automation and grid management
systems mainly have a master-slave structure,
because these applications are strictly hierarchically organized. In facility management clientserver networks are becoming popular. A client
polls the server (data point), or the server pushes the data periodically. In metering applications
both strictly hierarchical and client-server structures are used. All applications have in common
that the devices of the system represent the communication nodes of the PLC network, and each
of these nodes needs to be connected to the central node (master, server). This requirement is
challenging considering that many network
nodes are remote from the central node, and
perhaps serve time-critical applications like
those mentioned in the previous section. Furthermore, while individual nodes communicate
only small amounts of data at a time, the total
data volume to be transferred through the network is substantial. Hence, resource-efficient
transport of data to and from the central node is
mandatory to achieve sufficient network coverage.
108
Communications
IEEE
Robustness to Changes — In PLC networks
the communication channel may change abruptly
during normal operation. For example, switching
operations in medium-voltage energy systems to
balance the power consumption over the distribution grid will result in changes of channel
transfer functions in sizeable parts of the PLC
network. The PLC network design must be able
to cope with such abrupt changes, which means
that the connectivity must be maintained during
or quickly recovered after these changes. Since
severe network disruptions due to, say, physical
removal of network links, are often not exceptional events but occur frequently during normal
operation, maintenance of system availability is
only possible with redundant communication
links and autonomous use of redundancies. That
is, a PLC network that needs to estimate link
qualities and re-establish connections after
topology changes have occurred will not be able
to fulfill reliability requirements. Instead, ad hoc
networking features are needed. It is important
to note that the removal and addition of network nodes, or changes in the impedance of the
associated device also affect the communication
channels in a large neighborhood around this
node. This behavior is very different from wireless communications, where the mere presence
or absence of a wireless device does not affect
the link quality for another device.
Quality of Service — The main quality of service (QoS) requirements for the PLC network in
control and automation systems are high communication reliability, high overall network
throughput, and strict limits on delay. Often
messages transmitted from nodes to the central
node (e.g., notification about a sudden voltage
drop) or from the control center to network
nodes (e.g., a switching command for an actuator) are time-critical. Failure to meet delay
requirements can have serious consequences,
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
often with human safety at stake. In addition,
due to the time variance of the communication
channels, the functionality of all network nodes
needs to be verified continuously to guarantee
reliable data transfer and system response time.
Since optimization of throughput, reliability, and
delay often pose conflicting demands on the
design of the communication protocol, management of QoS requirements is a nontrivial task.
TRANSMISSION CONCEPTS FOR PLC
We now present transmission concepts apt to
meet the need for network coverage, link redundancies, and guaranteed QoS outlined above.
SINGLE-FREQUENCY NETWORK CONCEPT
The spatial dimension of the PLC network renders direct communication between the central
node and all other network devices infeasible.
To still achieve complete coverage, messages
need to be repeated, which is also known as
multihop transmission or relaying. Considering
that PLC reuses an existing infrastructure and
the broadcast nature of the PLC channel, an
altruistic repeater concept is appealing. That is,
network nodes that overhear a message for
another destination are prepared to retransmit
this message. Such a repeater concept makes
optimal use of the available communication
nodes in the network, and is flexible enough to
also ensure network coverage and communication reliability under changing channel conditions and topologies. In particular, the use of
multiple repeaters to relay the same message signal provides redundant signal paths, which are
needed to minimize network outages. To manage this multirelay transmission with minimum
use of communication resources, the single-frequency network (SFN) concept, which is known
from macrodiversity wireless communication systems [12], can be applied. The SFN allows all
repeaters to transmit simultaneously using the
same frequency band. The next receiving node(s)
sees a linear superposition of the retransmitted
signals, which is indistinguishable from a single
signal being sent over an equivalent multipath
channel. Hence, any communication technique
suitable for transmission over multipath channels
can be applied in an SFN. One popular method
is orthogonal frequency-division multiplexing
(OFDM) [5, 12]. We note that SFN-PLC transmission benefits from signal enhancement due to
concurrent retransmission. In some cases
destructive interference may occur, which can be
mitigated using distributed space-time coding
concepts presented in [8].
FLOODING CONCEPT
If altruistic relaying is used to route a message
through the network, the flooding concept is
implemented. We submit that flooding is an
attractive packet delivery process in control and
automation PLC networks for the following reasons. First, flooding eliminates almost all routing
overhead, which can be substantial for large networks with multiple repeater levels. Second, it is
extremely robust to network changes. This is
crucial for applications such as energy management or AGL systems. As mentioned above,
changes in PLC networks are often abrupt and
affect a large fraction of nodes. The need to
establish a route would compromise communication reliability and delay constraints. Third, considering the delivery of a single packet, flooding
minimizes the delay in that it always finds the
shortest path to the destination. Key for flooding
to be effective is the application of the SFN concept. SFN transmission avoids congestion for the
packet that is flooded, and thus accomplishes
efficient use of network redundancy and minimizes transmission delay.
On the downside, flooding has the potential
to create closed communication loops and to
massively occupy channel resources. Furthermore, different packets flooded simultaneously
in a specific geographical area can destroy each
other. To avoid or mitigate these effects, two
measures are suggested. First, active network
nodes (repeaters) monitor the packets or packet
numbers and ensure that every packet is repeated only once. Second, each packet is equipped
with a counter nrepeat that specifies the maximal
number of times a packet can be repeated before
it reaches the destination, and this counter is
decremented during each repetition. We note
that two different counters may be used for
transmission in opposite directions due to potentially non-reciprocal transfer functions or different interference situations at different locations.
IEEE
BEMaGS
F
The spatial
dimension of the PLC
network renders
direct communication between the
central node and all
other network
devices infeasible. To
still achieve complete
coverage, messages
need to be repeated,
which is also known
as multihop transmission or relaying.
MIXED DETERMINISTIC AND
RANDOM MAC CONCEPT
As discussed earlier, a feature common to many
applications is that traffic flows to and from a
central node, which suggests a centralized MAC
with a master-slave concept. The organization of
the transmission in the downlink direction (from
master to slaves) is simple since only the master
transmits data to one or multiple slaves.
The situation is different for the uplink,
where a number of slaves may have data to be
transmitted to the master at the same time. A
first option would be a purely deterministic
MAC protocol that completely eliminates signal
collisions. To this end, slaves could be successively polled by the master to see whether uplink
resources are required. To ensure a certain
polling rate, which defines the reaction time of
the PLC network, a certain fraction of channel
use has to be reserved for this mechanism. For
example, consider a sensor which monitors rare
events that occur on average, say, once every
day, but which require a fast response time of,
say, 10 s. The master would have to poll the sensor every 10 s, which is an excessive waste of
resources. Furthermore, while this approach may
be feasible in relatively small networks, the reaction time can quickly become unacceptably large
for networks with hundreds of nodes. For the
same reasons, other deterministic medium access
policies, such as master-slave-oriented bus protocols, token ring protocols, or solutions with fixed
time slots reserved for each individual network
node within a time-division multiple access
(TDMA) scheme are also not well suited. Random medium access techniques, such as Aloha
or carrier sense medium access (CSMA), offer
more flexibility in this regard. Since reliable car-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
109
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Figure 2. Tree topology as an example model for an energy distribution grid with 100 (left) and 200 (right) nodes (shown as circles). The
x- and y-axes are the main supply lines to which all nodes are connected. The master is located at the origin, and the slave nodes are
generated according to a uniform distribution over the diamond-shaped area defined by the maximum cable length between master and
slaves, where cables run in parallel to the x- and y-axes. The node density is kept constant regardless of the number of nodes.
rier sensing in PLC is complicated by the hidden
node problem [13], particular collision avoidance
techniques have to be employed like those developed for indoor PLC systems [14]. However,
application of mechanisms for solving the hidden
node problem and collision avoidance add considerable overhead for transmission with small
data packets and therefore are not well suited
for automation and control PLC systems. In case
of purely random access without collision avoidance (i.e., Aloha or slotted Aloha), the packet
delay can become unacceptably large, and network throughput is limited in the case of highly
loaded networks with a large number of nodes.
These considerations motivate a hybrid
MAC protocol that combines elements of deterministic and random medium access. First, the
master establishes a network-wide TDMA
frame structure through the broadcast of control packets to all nodes. Within each frame the
master allocates time slots for dedicated master-slave connections serving services such as
polling. Then the remaining time slots are used
for random medium access in the uplink, which
allows slaves to connect spontaneously with the
master (e.g., if an event that deserves quick
reaction from the master is detected or a slave
joins the network). For the sensor example
mentioned above, the master would poll the
sensor at regular intervals much larger than the
response time to ensure that the sensor is functional and synchronized to the TDMA frame
structure. When the sensor has an event to
report, it uses random access within the dedicated uplink frame to send the message. Considering that per-node link utilization is not
high, slotted Aloha or one of its variants is the
method of choice for the random access within
the TDMA frame structure. Since all network
resource control rests with the master, it dimensions the frame structure in accordance with
the QoS demands from the specific applications
served by the network. Furthermore, two mech-
110
Communications
IEEE
anisms due to SFN-based flooding work in
favor of random access with slotted Aloha.
First, as long as at least one repeater node
receives a signal originating from another node
successfully, the underlying message is not
annihilated even though signal collisions may
have occurred. Second, SFN transmission can
be exploited to reduce the waiting time between
retransmission attempts and thus improve overall transmission delay applying the concept of
local acknowledgments devised in [7].
REMARKS
We would like to remark on some of the challenges associated with the described SFN-based
flooding concept.
Synchronization — SFN-based flooding
requires a network-wide clock. This can be established through the TDMA structure, which is
maintained by broadcast packets sent by the
master. Every packet is equipped with a synchronization preamble [9] based on which a slave
adjusts its timing. Since broadcast packets serve
several purposes, they are sent regularly. Experiments have shown that timing synchronization
with an accuracy of very few (often fewer than
two) symbol intervals is easily achieved, which
also means that only very little additional guard
space between transmissions is required for an
SFN. We note that frequency synchronization is
not problematic due to the low carrier frequencies used in narrowband PLC.
Channel Estimation and Error Propagation
— The need for receiver-side channel estimation
in SFN-based flooding can be bypassed using
differential modulation (e.g., across subcarriers
if OFDM is applied). Likewise, the effect of
error propagation can be neglected assuming the
use of error detection, such that only nodes
which deem a packet as received correctly will
retransmit it.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Channel Occupation and Energy — Clearly,
there is a price to be paid for not needing a
route. First, redundant retransmissions occupy
channels unnecessarily and prevent other messages being sent. When one TDMA frame travels, no other frame can travel in a certain
geographical area, which means that spatial
reuse and therefore packet origination rates are
limited. Second, the use of many retransmissions
wastes energy, which undermines the purpose of
smart grids. These problems are mitigated
through flooding using counters as described
above. Variations of flooding that aim at reducing the number of redundant broadcasts [15] can
further remedy the situation. Since the location
of many network devices is static, topology information may be used during flooding. For example, in parts of the network with high
connectivity, only a subset of nodes relays a message. Furthermore, signal waves can be directed
toward the destination, which also increases multiplexing capability. However, topology information needs to be exploited with care as, for
example, in ring topologies found in mediumvoltage energy distribution grids or AGLs (Fig.
1), a single node failure can completely change
the direction in which the signal wave needs to
propagate to reach the destination.
NUMERICAL RESULTS:
FLOODING VS. ROUTING
To make the argument for SFN-based flooding
more concrete, in this section we provide numerical performance results for three specific criteria pertinent to large-scale PLC networks. We
compare flooding with the alternative of centralized proactive routing, for which every message
is only retransmitted by exactly one repeater
node at every repetition level, and the route is
determined at the master node based on link
quality information given by the packet error
rate (PER).
AVERAGE DURATION OF A POLLING CYCLE
Polling is often one of the fundamental network
operations. For example, in AGL automation it
is critical to continuously monitor the functionality of all devices; thus, the master frequently
polls all slaves.
—
The average duration of the polling cycle D is
defined as the average time the master needs to
complete a single packet-request-response service with every slave. An
analytical expression
—
for a lower bound for D for the case of routing
systems can be obtained by making the idealized
assumption that the instantaneous PERs for all
node-to-node links are available at the—master.
To obtain an analytical expression for D for the
flooding-based system, we assume that the maximal number of packet
retransmissions, nrepeat, is
—
chosen such that D is minimized. Furthermore,
we have simulated SFN-based flooding using an
adaptation method for nrepeat.
—
Table 1 shows the numerical
results for D
—
routes, Dflood,opt assumrout,opt assuming optimal—
ing optimal nrepeat, and Dflood,adapt with adapted
n repeat for a ring topology (Fig. 1) with 10 and
100 uniformly distributed nodes, and for a ran-
IEEE
BEMaGS
Ring topology
F
Tree topology
Channel model
—
D rout,opt
—
D flood,opt (ES)
—
D flood,opt (EC)
—
D flood,adapt (ES)
—
D flood,adapt (EC)
10 nodes
100 nodes
100 nodes
200 nodes
30
427
421
1027
29
28
419
403
387
367
993
939
30
29
423
404
393
368
1003
945
Table 1. Comparison of the average duration of a polling cycle using optimal
routing and flooding with optimal and adapted number of retransmissions
nrepeat. Duration is measured in number of time slots. EC refers to the case
when the energies of the multiple signal paths are aggregated at the receiver; in
ES only the individual channel with the largest energy is considered.
domly generated tree topology with 100 and 200
nodes, respectively. The latter are illustrated in
Fig. 2. For SFN-based flooding we distinguish
two scenarios for superposition of simultaneously relayed signals: aggregation of signal energies
from all signal paths (denoted energy combining
[EC] in Table 1) and selection of the individual
channel with the largest energy (denoted energy
selection [ES]).
The figures in Table 1 are the average duration of a polling cycle measured in number of
time slots. We observe that flooding consistently
achieves a lower polling cycle duration than
routing. This is due to the fact that with flooding
the packet is received via the optimal route and
via additional repeater paths, and thus the probability of successful transmission is increased
compared to routing. Furthermore, the restriction of the number of repetitions in flooding due
to nrepeat avoids unnecessary occupation of channel resources. We further observe that performance degradations due to in situ adaptation of
nrepeat are less than 4 percent.
DURATION OF A BROADCAST TRANSMISSION
In the considered PLC networks messages are
frequently broadcast from the master to all
slaves. The purpose of broadcast messages is
manifold. It serves, for example, to update the
TDMA frame structure, to inform slaves of
which slots are used for specific services and
physical-layer parameters, to download software
updates, or to transmit fast time-varying application-specific parameters such as the distance of
an approaching airplane in AGL automation systems.
Flooding is a natural fit for fast broadcast
transmission as it utilizes the very broadcasting
nature of the PLC channel. The duration of a
broadcast is simply max{nrepeat} times the duration for a downlink slot. In case of routing, we
assume that the master sends the broadcast message to a number of slaves selected such that the
union of all nodes that receive the message
transmitted along those routes forms the complete set of nodes. In this way, the address field
of the broadcast packet is not expanded compared to a unicast packet. To minimize the dura-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
111
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Ring topology
Tree topology
Channel model
10 nodes
100 nodes
100 nodes
200 nodes
Routing
5
8
37
73
Flooding
3
4
5
5
Table 2. Comparison of the average duration for a broadcast to all slaves using
routing and flooding. Duration is measured in number of time slots.
ES case
EC case
800
Duration of polling cycle (number of slots)
750
700
650
600
550
500
450
400
350
-15
-10
-5
0
5
10
15
20
25
30
Polling cycle (0 = switch of channel mode)
Figure 3. Duration of polling cycles for flooding before and after an abrupt
topology change.
tion of the broadcast, we apply a greedy algorithm that selects the next slave such that a maximal number of nodes is reached along its route.
Table 2 shows the time needed to deliver a
broadcast message with routing and flooding for
the same four network topologies considered for
polling. The time is measured in number of time
slots, and the particular figures were determined
such that the PER for all slaves is less than 0.1
percent. It is interesting to observe that performance differences between flooding and routing
are moderate for ring structures, whereas they
become significant for tree topologies. In particular, flooding is rather insensitive to the underlying topology since signal waves propagate in all
directions. Likewise, the actual number of nodes
is insignificant as long as the spatial extension of
the network does not change significantly.
ROBUSTNESS TO TOPOLOGY AND
CHANNEL FLUCTUATIONS
We have already pointed out earlier that the
quality of communication links in a PLC network can vary with time due to, say, load
changes. Switching operations even change the
network topology. A typical example of a severe
topology change would be the opening of a
medium-voltage distribution line ring structure
at a location close to the master node.
112
Communications
IEEE
A
BEMaGS
F
To illustrate the agility of flooding, we simulated polling cycles for 100 nodes that first were
arranged in a ring structure and after a certain
number of cycles were rearranged into a random-tree structure. We hasten to say that such
a dramatic topology change is unrealistic and
only serves as an extreme academic example to
test the robustness of flooding. In the simulations a request to a slave was repeated until the
master successfully received a response. Figure
3 shows the measured duration for a polling
cycle, where the network topology is changed
before cycle number 1. We observe that before
the topology change, the duration of the polling
cycle jitters around average values of about 410
(EC case) and 425 (ES case) time slots. The
length of the polling cycle jumps to about 700
(EC) and 780 (ES) slots right after the network
change, but it is already reduced again by about
250 slots in the following cycle. Already after 5
to 7 polling cycles, the cycle durations have converged to the (new) stationary values. To have
an estimate for the length of the adaptation
process in absolute time, we assume a slot duration of 10 ms. This value is typical for PLC
transmission in the CENELEC-A band with a
bandwidth of 50 kHz. Then the average duration of a polling cycle for 99 slaves would be
between 4 and 4.5 s according to the results in
Fig. 3. The first polling cycle after the abrupt
topology change would require 7 to 8 s to successfully reach all slaves, but within only 30 to
40 s the adaptation to the new topology is completed. This very fast adaptation, which is only
possible with an algorithm with short memory,
satisfies real-time requirements for the communication system in, say, Smart Grid applications
even during and after a switching instant. Hence,
SFN-based flooding is a very attractive solution
in this regard as well.
CONCLUSIONS
In this article we have presented an overview of
the requirements for PLC to become an enabler
for advanced control and automation systems
such as energy management and facility automation systems. Starting from these requirements
we have described suitable PLC transmission
concepts. We have advocated, and in part substantiated with numerical evidence, that the
combination of single-frequency networking
with flooding embedded into a hybrid MAC
protocol is attractive to meet the application
requirements. The efficacy of the presented
concepts has been verified in field trials under
the umbrella of the REMPLI project, and PLC
products based on this technology are currently
being used in a number of pilot projects for
advanced meter management (e.g., in Karczew,
Poland) and streetlight control (e.g., in Fürth,
Germany). We close by noting that the future
proliferation of PLC as an enabler for Smart
Grid functionalities is not only a technological
issue, but also depends strongly on how swiftly
electric utilities are able to implement necessary
changes in the energy distribution process, and
ongoing legislative developments concerning
infrastructure reliability and new services like
real-time pricing.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
REFERENCES
BIOGRAPHIES
[1] H. Ferreira et al., “Power Line Communication,” in Encyclopedia of Electrical and Electronics Engineering, J.
Webster, Ed., Wiley, 1999, pp. 706–16.
[2] H. A. Latchman and L. W. Yonge, “Power Line Local
Area Networking,” IEEE Commun. Mag., vol. 31, no. 4,
Apr. 2003.
[3] S. Galli, A. Scaglione, and K. Dostert, “Broadband Is
Power: Internet Access through the Power Line Network,” IEEE Commun. Mag., vol. 31, no. 5, May 2003.
[4] S. Galli and O. Logvinov, “Recent Developments in the
Standardization of Power Line Communications within
the IEEE,” IEEE Commun. Mag., vol. 46, no. 7, July
2008, pp. 64–71.
[5] G. Bumiller, “Single Frequency Network Technology for
Medium Access and Network Management,” IEEE
ISPLC, Athens, Greece, Mar. 2002.
[6] G. Bumiller, “Power-Line Physical Layer Emulator for
Protocol Development,” IEEE ISPLC, Zaragossa, Spain,
Mar. 2004.
[7] L. Do, H. Hrasnica, and G. Bumiller, “SALA MAC Protocol for PLC Networks Based on Single Frequency Network Technique,” IEEE ISPLC, Orlando, FL, Mar. 2006.
[8] L. Lampe, R. Schober, and S. Yiu, “Distributed SpaceTime Block Coding for Multihop Transmission in Power
Line Communication Networks,” IEEE JSAC, vol. 24, July
2006, pp. 1389–1400.
[9] G. Bumiller and L. Lampe, “Fast Burst Synchronization
for Power Line Communication Systems,” EURASIP J.
Adv. Signal Process., vol. 2007, article ID: 12145.
[10] A. Scaglione and Y.-W. Hong, “Opportunistic Large
Arrays: Cooperative Transmission in Wireless Multihop
Ad hoc Networks to Reach Far Distances,” IEEE Trans.
Signal Process., vol. 51, no. 8, Aug. 2003, pp. 2082–92.
[11] E. Garrity, “Getting Smart,” IEEE Power Energy Mag.,
vol. 9, Mar./Apr. 2008, pp. 38–45.
[12] M. Eriksson, “Dynamic Single Frequency Networks,”
IEEE JSAC, vol. 19, Oct. 2001, pp. 1905–14.
[13] M. Mushkin, “A Novel Distributed Synchronized Media
Access Control Mechanism and Its Applicability to InHouse Power-Line Networking,” IEEE ISPLC, Malmö,
Sweden, Mar. 2001.
[14] M. Lee et al., “HomePlug 1.0 Powerline Communication LANs — Protocol Description and Performance
Results,” Int’l. J. Commun. Sys., vol. 46, no. 5, June
2003, pp. 447–73.
[15] T. Kwon et al., “Efficient Flooding with Passive Clustering — An Overhead-Free Selective Forward Mechanism
for Ad Hoc/Sensor Networks,” Proc. IEEE, vol. 91, no. 8,
Aug. 2003, pp. 1210–20.
GERD BUMILLER [M] (gerd.bumiller@iad-de.com)
_______________ received his
Diplom (Univ.) and Ph.D. degrees in electrical engineering
from the University of Erlangen, Germany, in 1997 and
2009, respectively. He joined iAd GmbH, Germany, as a
communication system developer in 1997. Since 2000 he
has been chief developer responsible for all power line
communication projects of iAd. In this role he has been
involved in a number of European and international projects on high-data-rate access, control and automation,
and Smart Grid power line communications. He has published widely in the area of power line communications
with contributions to channel measurement and modeling,
coupling, synchronization, multicarrier modulation, coding,
routing, and ad hoc networking. He is a member of Work
Group 2 (Network Operations) of the EU Smart Grids initiative (www.smartgrids.eu), participates in the Open Metering initiative of ZVEI and FIGAWA in Germany, and is a
member of AK 461.0.141 and UK 716.1 in the German
standardization body DKE.
IEEE
BEMaGS
LUTZ LAMPE [SM] (Lampe@ece.ubc.ca)
___________ received his Diplom
(Univ.) and Ph.D. degrees in electrical engineering from the
University of Erlangen, Germany, in 1998 and 2002,
respectively. Since 2003 he has been with the Department
of Electrical and Computer Engineering at the University of
British Columbia, where he is currently an associate professor. His research interests are in communication theory
with applications to wireless and power line communications. He is Vice-Chair of the IEEE Communications Society
Technical Committee on Power Line Communications. He
was General Chair of the 2005 International Symposium on
Power Line Communications and the 2009 IEEE International Conference on Ultra-Wideband.
F
The efficacy of the
presented concepts
has been verified in
field trials under the
umbrella of the
REMPLI project, and
PLC products based
on this technology
are currently being
used in a number of
pilot projects for
advanced meter
management.
HALID HRASNICA (hrasnica@eurescom.eu)
_____________ graduated in 1993
in electrical engineering from the University of Sarajevo,
Bosnia and Herzegovina. Afterward, he worked in Energoinvest Communications in Sarajevo as developing engineer for communications systems. In 1995 he joined the
Institute for Telecommunications at Dresden University of
Technology, Germany, where he received his Ph.D. degree
in electrical engineering and information technology. Since
February 2006 he has been with Eurescom GmbH in Heidelberg, Germany, where he works as programme manager for research projects on future telecommunications
networks.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
113
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
ACCEPTED FROM OPEN CALL
IMS-Compliant Management of
Vertical Handoffs for
Mobile Multimedia Session Continuity
Paolo Bellavista, Antonio Corradi, and Luca Foschini, University of Bologna
ABSTRACT
The recent advances in wireless client devices
and the crucial role of multimedia communications in our society have motivated relevant standardization efforts, such as the IP multimedia
subsystem, to support session control, mobility,
and interoperability in all-IP next-generation
networks. IMS has already driven the design of
commercial mobile multimedia, but exhibits limited support for service continuity during handoffs. In particular, it omits advanced techniques
to reduce/eliminate handoff delays, especially
during vertical handoffs (i.e., change of the wireless technology employed by a client to access
the wired Internet, e.g., from UMTS to WiFi).
We propose an original solution for session continuity based on the primary design guideline of
cleanly and effectively separating the signaling
plane (for session reconfiguration via SIP) from
the media delivery plane (data transmission and
related handoff management operations). Our
optimized handoff management techniques
exploit terminal-based decentralized predictions
to minimize service-level handoff delays. Different from other recent related work, our proposal
fully complies with the standard IMS infrastructure and works at the application level. The
reported experimental results point out that our
solution, available as an open source tool for the
IMS community, reduces playout interruption
times relevantly by introducing a limited and
scalable signaling overhead.
INTRODUCTION
1
Additional information,
experimental results, and
the IHMAS prototype
code are available at
http://lia.deis.unibo.it/Res
earch/IHMAS/
________
114
Communications
IEEE
A growing number of mobile users require seamless access to multimedia services, such as
audio/video streaming, while they move across
the available, possibly heterogeneous, wireless
infrastructures, spanning from IEEE 802.11
(WiFi) and Bluetooth (BT) to cellular 3G. However, despite the great potential of multimedia
applications over integrated and heterogeneous
wired-wireless networks, their development and
deployment are still challenging tasks due to typically strict QoS requirements (e.g., data arrival
time, jitter, and data losses). In addition, user
mobility introduces demanding issues such as
0163-6804/10/$25.00 © 2010 IEEE
bandwidth fluctuations and/or temporary loss of
connectivity due to device handoffs, when a
mobile node (MN) disconnects from one access
point (AP) and reconnects to a new one. In particular, granting service continuity during vertical
handoffs is one of the open and crucial problems
still to overcome, by defining adequate and
effective support for interoperable session management.
A large group of standardization entities,
which range from the Third Generation Partnership Project (3GPP) to the Internet Engineering
Task Force (IETF) and Open Mobile Alliance
(OMA), has recently specified the IP multimedia
subsystem (IMS) [1]. IMS defines an overlay
architecture for session control and authentication, authorization, and accounting (AAA) in
all-IP next-generation networks. The main goal
is achieving openness and interoperability via an
application-layer approach, mainly by exploiting
the Session Initiation Protocol (SIP). At its current stage, however, IMS exhibits some limitations in the support of handoff management. In
particular, IMS adopts a reactive approach and
starts multimedia session reconfiguration only
after losing connection with origin APs, thus
postponing the execution of data handoff management operations, which may be articulated,
especially in the case of vertical handoffs. That is
prone to produce significant handoff delays and
compromise session continuity.
The article tackles this problem by proposing
a novel solution that exhibits three original characteristics. First, it decouples session signaling
and data management, by exploiting decentralized handoff predictions that proactively activate
session signaling with the new target network
while data still flow over the old AP. Second, it
is fully compliant with the IMS standard and
does not require any change to already installed
IMS equipment. Third, it applies optimized techniques for data management during handoffs to
minimize data losses. The proposed support
infrastructure has been implemented as an open
source tool, called the IMS-compliant Handoff
Management Application Server (IHMAS); 1
IHMAS is available for the IMS community, and
outperforms related work in the field in terms of
handoff delays and scalability [2–4].
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
The article is structured as follows. The next
section briefly overviews the IMS background and
related work. Then we motivate the need for
proactive handoff management and propose different techniques to optimize IMS-based handoff.
We then present the IHMAS design, implementation, and experimental results, while directions of
our current research conclude the article.
SESSION CONTINUITY IN IMS:
BACKGROUND AND RELATED WORK
The IMS standard supports different kinds of
mobility. Our proposal focuses on terminal
mobility management during service provisioning
and, in particular, on vertical handoffs. Hence,
in the following we use the term session continuity to indicate specifically multimedia service
continuity in the case of so-called mid-call terminal mobility with vertical handoffs.
IMS BACKGROUND
IMS allows the creation, modification, and termination of service sessions independent of
underlying data link layer technologies and
transport protocols. In particular, IMS promotes
the clear separation of the signaling plane — for
session re-/configuration, based on all-over-IP
technologies and SIP — and the media delivery
plane — for data transmission, based on different possible transport protocols, such as IETF
Real-Time Transport Protocol over UDP (RTPover-UDP).
In addition, the IMS infrastructure offers several facilities to help multimedia service developers. It not only defines session and AAA
protocols, but also provides a wide set of support
functions (e.g., for decentralized proxy-based
session management, endpoint localization, and
introduction of new IMS-based services/extensions [1]). The core IMS functional entities are:
• IMS client, which controls session setup and
media transport by implementing all the
SIP extensions specified by IETF and 3GPP
IMS-related standards. A unique HTTPlike universal resource identifier (URI),
such as sip:user@domain, identifies each
IMS client. Any session is set up between
two IMS clients (SIP endpoints). In this
article, without losing any generality and for
the sake of presentation clarity, we always
consider an MN as the originating SIP endpoint and a fixed correspondent node (CN)
as the terminating one.
• Proxy-call session control function (PCSCF), which establishes secure associations
with
MNs
and
routes
outgoing/incoming SIP messages to the
inner IMS infrastructure on an MN’s behalf.
• Interrogating-CSCF (I-CSCF), which is
responsible for securely interconnecting and
routing SIP messages among different IMS
domains.
• Application server (AS), which allows the
introduction of new IMS services and extensions by having full control over traversing
SIP dialogs.
• Home subscriber server (HSS), which stores
client AAA data and profiles.
• Serving-CSCF (S-CSCF), which is the most
important session control component and
enables the coordinated interaction of all
IMS entities. The S-CSCF receives register
requests from IMS clients, and interacts
with HSSs to authenticate them, and to
update associations between SIP URIs and
current client endpoints. Moreover,
depending on filters/triggers specified by
client profiles, the S-CSCF may either route
incoming SIP messages directly to a CN —
through the terminating P-/I-CSCF if a CN
is in the same/different domain — or forward them to an AS [1].
• Dynamic Host Configuration Protocol
(DHCP), the standard Internet facility that
IMS employs for MN configuration.
Given its ability to change SIP message content, the S-CSCF can also extend MN-to-CN
session signaling paths through the interposition
of convenient ASs; these ASs may participate in
multimedia content adaptation/transport/buffering by interacting with multimedia proxy (MP)
entities, which play the role of media gateways at
the media delivery plane. ASs are the only points
of contact between signaling and data transport
planes, and can enforce data handoff management operations by controlling MPs along MNto-CN data paths. The IMS client executes at
the MN; DHCP, P-CSCF, and MPs are deployed
at the network edges of the MN visited networks; ASs, HSSs, I-CSCFs, and S-CSCFs typically run in home networks.
With a closer view to detail, IMS-based vertical handoff management includes several functions at different protocol stack layers. At the
data link layer, the crucial operation is data link
connection, which includes monitoring origin AP
connection loss and establishing a new connection with a target AP. The main network-layer
operation is the renewal of network configuration parameters at an MN: according to the
standard IMS specification, IHMAS exploits
DHCP to re-establish the MN IP address and PCSCF server endpoint in the target network.
Afterward, application-layer session signaling
starts. First, an MN establishes a new secure
association with its S-CSCF by sending a REGISTER message through the target P-CSCF. Let
us note that this is one of the longest handoff
management operations, requiring two roundtrips to fulfill IMS security requirements, and
lasts about 320 ms [1]. Then the MN renegotiates the ongoing session by sending an INVITE
message (about 85 ms), with updated Session
Description Protocol parameters, such as RTPover-UDP MN endpoints and a network description. The INVITE message triggers data flow
rebinding (and content adaptation when necessary) at the media delivery plane.
Figure 1 shows the standard message flow
during IMS-managed handoffs; our original proposal is to enhance it via AS interposition, as
detailed in the following. Continuous black lines
over grey background represent the IMS session
signaling protocol, dotted black lines the local
events emitted by the MN (e.g., old connection
loss), and solid dark blue ones the data streams.
Let us remark that IMS handoff may incur relatively long delays (up to 2 s) due to its reactive
IEEE
BEMaGS
F
The main networklayer operation is the
renewal of network
configuration
parameters at MN:
according to the
standard IMS
specification, IHMAS
exploits DHCP to
re-establish MN IP
address and P-CSCF
server endpoint in
the target network.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
115
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
MN
Old conn.
loss
Target P-/I-CSCFMN
S-CSCFMN
S-CSCFCN
CN
Data flow
Media loss
(4) 401 Unauth.
(5) REGISTER
(8) 200 OK
(9) INVITE
(2) REGISTER
(3) 401 Unauth.
(6) REGISTER
(7) 200 OK
(10) INVITE
(11) INVITE
(12) INVITE
Data flow
rebind
(16) 200 OK
(15) 200 OK
(14) 200 OK
(13) 200 OK
Renegotiation
phase
(1) REGISTER
Registration
phase
Datalink
handoff
and
DHCP req.
Data flow
Figure 1. Message flow for non-optimized IMS handoff management.
approach: handoff is triggered by the loss of the
old connection, and all the subsequent operations are performed sequentially, without exploiting the possible overlapping between the
coverage areas of the old and target APs.
RELATED WORK
Application-layer IMS-based handoff management has its roots in SIP-based mobility management, first proposed by Schulzrinne [5].
Thereafter, a number of SIP-based research
efforts tackled session continuity by addressing
two main issues:
• Session reconfiguration, usually managed at
the application layer with SIP
• Seamless flow provisioning and data path
rebinding, managed at either the application layer through SIP rebinding or, more
often, the network layer via Mobile IP
(MIP) and its extensions [6]
The interest in IMS-based session continuity
is also demonstrated by the recent special issue
of this magazine on IMS infrastructures and services [7]. In the following, for the sake of briefness, we only focus on the three vertical handoff
solutions closest to IHMAS, ordered by growing
similarity.
Intelligent Network-Seamless Mobility Access
(IN-SMA) mainly focuses on session reconfiguration [2]. In particular, IN-SMA introduces an
IMS AS, called the mobility server, that supports
voice call (re)routing from 3G cellular networks
to IMS-compliant WiFi VoIP infrastructures,
while providing seamless flow provisioning via
MIP. Like our approach, IN-SMA considers
mobility as an advanced IMS service (implemented as an IMS AS) and does not require
changes to the IMS core. However, it presents
two main limitations. First, IN-SMA supports
vertical handoffs only in the 3G-to-WiFi direction, and excludes decentralized, terminal-based,
and proactive MN initiation of handoff management countermeasures. Second, the proposed
solution for flow continuity seems to still be pre-
116
Communications
IEEE
A
BEMaGS
F
liminary: the authors propose MIP usage but do
not face at all the challenging issues of realworld handoff latencies.
Networking Context-Aware Policy Environment (NetCAPE) also supports 3G-to-WiFi vertical handoffs. It adopts a cellular-centric
solution by employing the so-called tight-coupling interworking scheme of 3GPP [3]. Similar
to our proposal, NetCAPE exploits handoff prediction to proactively activate vertical handoff
procedures before clients lose connectivity from
their origin APs. However, its tight-coupling
approach requires that any communication from
the WiFi domain passes through the core cellular network, with increased communication
costs. Moreover, NetCAPE focuses on handoff
prediction and MIP delay optimization, but not
on full compliance with the standard IMS infrastructure.
To the best of our knowledge, the published
solution closest to IHMAS is [4], which enables
secure session continuity for the multimedia
domain (MMD), the IMS-equivalent infrastructure defined by the 3GPP2 standard. MMD
aims to reduce handoff latency by employing
context transfer techniques to move, either reactively or proactively, session description and
security information from old to new P-CSCFs,
and addresses all MIP-related security issues [4].
Nonetheless, it is not easily deployable in the
IMS infrastructure: it requires modifying the
standard IMS session signaling protocol to support session context transfer and consistently
changing existing P-CSCFs. In addition, MMD
operates data handoff (MIP address reconfiguration) before multimedia session renegotiation,
thus precluding multimedia flow adaptation
before data transfer. Finally, the proposed prototype employs MIPv4, thus possibly undertaking
long latencies due to triangular routing and, consequently, long handoff delays [4].
IHMAS APPLICATION-LAYER
HANDOFF ENHANCEMENTS
As exemplified in the previous section, providing
session continuity of IMS-based mobile multimedia is still an open issue. About session signaling,
one of the most challenging problems is to
reduce the duration of session rebinding. In fact,
the rebinding of ongoing sessions requires
exchanging several SIP REGISTER/INVITE messages (usually called re-REGISTER and reINVITE), thus lengthening data handoff delays.
Some support solutions in the literature face this
issue by introducing extensions to deal with context transfer [4]. However, these solutions are
not standard; their deployment requires protocol
stack changes at all CSCFs, and their main goal
is not session continuity. For instance, MMD
does not allow data adaptation before data
handoff, and this may cause the injection into
the target network of multimedia flows with
quality exceeding its transmission capabilities,
thus possibly congesting it up to data delivery
interruption.
About data handoff management, current
proposals tend to be MIP-based and with still
unsolved issues. First, the MIP network layer
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
approach lacks the visibility of (and hardly complies with) application-specific requirements
such as maximum tolerable delay. Moreover,
there is a growing interest in soft data handoff
strategies because they permit MNs to have two
or more simultaneous connections to various
APs, as opposed to hard strategies where MNs
have only one connection at a time. MIP-based
solutions typically adopt a hard handoff strategy
that maps one MN to only one IP address, by
excluding more flexible rebinding opportunities.
Finally, for each vertical handoff, MIP and
MIPv4 require an additional round-trip time (in
addition to the ones due to REGISTER/INVITE
messages) to reconfigure data endpoints at the
MN home (i.e., home agents).
Our IHMAS proposal overcomes all the
above limitations by adopting three main design
guidelines. First, our solution is fully IMS-compliant: IHMAS originally exploits IMS separation of signaling/media delivery planes and
introduces a new AS to realize advanced data
handoff management. In that way, the proposal
is fully compatible with the current IMS standard, without the need for any intervention on
already deployed IMS components. Second, our
solution is proactive: to reduce vertical handoff
latency, IHMAS predicts vertical handoffs at
clients and starts session reconfiguration before
handoff management. Third, our proposal adopts
an application layer proxy-based approach, not
only for session signaling, but also for data handoff. IHMAS overcomes typical MIP problems by
using SIP-based flow rebinding and exploiting
MPs that are locally deployed at client access
localities to enable prompt, flexible applicationlayer handoff management on the media delivery plane.
By focusing on session signaling, proactivity is
based on the introduction of a vertical handoff
predictor (VHP) that extends the IMS client by
providing lightweight and completely decentralized prediction via only local access to client
wireless interfaces. The IMS client exploits VHP
predictions to proactively start SIP-based session
reconfiguration (REGISTER/INVITE messages)
over the target AP while continuing to receive
multimedia data from the origin AP. On the
IMS infrastructure side, the primary IHMAS
component is the AS for session continuity
(ASSC) that reduces handoff media losses by
decoupling session rebinding and data transfer
times. Deployed at the MN home network, the
ASSC receives all SIP messages from the MN SCSCF; it acts as an IMS anchor point, by hiding
CN from MN mobility and consuming all handoff-related SIP messages (especially INVITE);
finally, it triggers data handoff control operations over the MP.
The MP component implements our novel
handoff management strategies for either soft or
hard (the latter either reactive or proactive)
handoff management. In particular, we propose
a flexible two-level buffering solution: the MP,
deployed along the MN-to-CN multimedia data
path, hosts one second-level buffer for each MN
currently active in its domain to suitably adjust
multimedia flows with the main goal of service
continuity [6]. The second-level buffer enables
soft handoff management by locally supporting
the duplication (and simultaneous transmission)
of multimedia flows over multiple wireless interfaces in the last wired-wireless hop. For hard
proactive handoff, instead, the second-level
buffer can receive and store all the incoming
packets that would be lost during the (short)
RTP-over-UDP rebinding period and enable
local retransmission of stored packets (forwarding) when the MN reconnects to the MP over
the target AP. The hard reactive handoff strategy is a simplified case of hard proactive with no
data store-and-forward. Finally, let us observe
that in the following, we focus only on the downlink direction (from APs to wireless clients)
because most traffic load is in this direction in
multimedia streaming applications such as video
on demand and IP-based TV. Anyway, the proposed IHMAS optimizations are lightweight and
may easily also apply to uplink data transmissions by only requiring limited buffering
resources at the client side (i.e., to cover only
the handoff disconnection interval), usually
available on any portable client device nowadays.
The remainder of the section presents our
original IMS handoff optimizations. While all
IHMAS enhancements adopt a proactive
approach on the IMS session signaling plane, we
distinguish different types of IHMAS improvements depending on how the ASSC exploits its
awareness of vertical handoff temporal proximity
— through (re-)INVITE reception from MN —
to trigger different management actions at the
media delivery plane: hard reactive/proactive
and soft data handoff management.
IEEE
BEMaGS
F
The proposed
IHMAS optimizations
are lightweight and
may easily apply also
to uplink data
transmissions by only
requiring limited
buffering resources
at the client side,
i.e., to cover only the
handoff disconnection interval, usually
available at any
portable client device
nowadays.
IHMAS HARD DATA HANDOFF MANAGEMENT
Figure 2 shows our optimized hard
proactive/reactive handoff procedures. Dashed
black lines represent our original message flow
extensions and red brackets point out the operations that the MP executes only for proactive
management. VHP predictions enable the proactive execution of data link/network handoff
operations over the target network: they permit
completing the registration phase and starting
renegotiation while keeping the media flows
active over old connections. In particular, upon
INVITE message reception (step 12), depending
on the adopted handoff strategy, the ASSC triggers data handoff at the MP by either activating
(proactive) or not (reactive) the store-and-forward function on multimedia flows (step 13).
For proactive handoff, the MP receives and
stores incoming flows in its local second-level
buffer during the RTP-over-UDP rebinding period, while the IMS client supports session continuity by consuming the client-side buffer. Thus,
after data rebinding, the MP can promptly fill up
MN buffers by retransmitting all the data otherwise lost due to temporary disconnection.
Therefore, hard proactive management
requires that the MN sends a message to trigger
data retransmission. RTP retransmissions were
standardized only recently and some IMS clients
do not yet support this facility, as well as RTP
data frame reordering to present retransmitted
frames in the correct order to multimedia players [8]. To guarantee the widest interoperability,
in addition to the more effective hard proactive
strategy, IHMAS provides hard reactive handoff.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
117
A
BEMaGS
F
Communications
IEEE
MN
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Target P-/I-CSCFMN
MP
S-CSCFMN
ASSC
CN
Handoff
prediction
Datalink handoff
and DHCP req.
(1) REGISTER
(2) REGISTER
(3) 401 Unauth.
(4) 401 Unauth.
(5) REGISTER
(6) REGISTER
(8) 200 OK
(9) INVITE
Media loss for
reactive mng.
Old
conn.
loss
Data flow
(17) 200 OK
Lost frames
Data flow
(7) 200 OK
(10) INVITE
(11) INVITE (12) INVITE
(13) Hard
(proactive)/
reactive
data handoff
trigger
Proactive data buffering
Data flow rebind
(14) 200 OK
(16) 200 OK (15) 200 OK
Lost data retransm.
(18) Client
re-transmit
request
Figure 2. Message flow for optimized hard reactive/proactive data handoff
management.
Hard reactive handoff may also be useful, in
some cases, to minimize the exploitation of second-level buffer resources. However, that buffering cost is usually very low; in fact, it depends on
single multimedia frame size, data rate, and
especially RTP-over-UDP rebinding time, usually below 220 ms even in high-load scenarios such
as the ones discussed below. In addition, let us
note that hard handoff strategies should apply
only if compatible with application-specific
requirements on delay and data loss; otherwise,
soft handoff has to be selected.
Our proposal does not require any modification to the standard IMS protocol (grey background), but only adds some new messages
(steps 13 and 18) to link signaling session and
media delivery planes. Different from MMD, we
do not postpone session renegotiation after data
handoff; in this way, IHMAS can tailor multimedia flows to target network characteristics before
transferring them, by granting session continuity
and flow delivery at the most suitable quality at
any time [4]. To activate flow adaptation at the
MP proactively, the ASSC exploits Session
Description Protocol data in INVITE messages.
IHMAS SOFT DATA HANDOFF MANAGEMENT
Figure 3 shows the optimized soft handoff procedure implemented by IHMAS. Compared to
hard handoff, soft handoff supports a wider
range of multimedia services, even with stricter
delays and jitter requirements. It requires smaller second-level buffer resources, only to store
the short streaming data chunk necessary to sustain streaming; for instance, for a G.711 (U-law)
encoded audio flow with constant frame rate =
50 frame/s and G.711 RTP packet payload size
of 216 bytes, it is sufficient to store three data
frames (less than 1 kbyte). In addition, it slightly
increases energy consumption at the client by
118
Communications
IEEE
A
BEMaGS
F
maintaining the old interface active for only the
short time necessary to receive the first packets
over the new interface, usually below 220 ms.
IHMAS soft handoff adopts a proactive session
signaling approach and is completely IMS-compliant: it solely requires minor modifications at
IMS clients, only affecting media delivery modules.
By delving into finer detail, at INVITE message reception, the ASSC activates soft handoff
at the MP (step 13). Hence, the MP duplicates
and sends RTP frames over both old and target
wireless links. The IMS client is in charge of
removing duplicated frames. The interaction
ends at the completion of RTP-over-UDP connection rebinding. To stop data flow duplication
at the MP, IHMAS introduces a new control
message from MN to MP by adding an optional
field to the standard RTP retransmission payload format (step 18) [8].
IHMAS TESTBED IMPLEMENTATION
AND PERFORMANCE RESULTS
To thoroughly test and evaluate the performance
of IHMAS, we carried out two types of experiments by deploying it in the widescale and heterogeneous wireless network at our campus.
First, we performed field experiments validating
the effectiveness of IHMAS functions. Second,
we used a state-of-the-art IMS traffic generator
to assess the cost and scalability of the proposed
signaling by emulating the typical behavior
expected in wide-area IMS deployments with a
multitude of concurrent users/requests.
In the evaluation testbed, the IMS infrastructure components (CSCFs, HSSs, and
ASSC) run on Linux boxes with two 1.8 GHz
CPUs and 2048 Mbytes RAM, following the
standard IMS deployment indicated earlier.
During the first experiment, MNs (Windows
and Linux laptops equipped with OrinocoGold
WiFi cards and MopogoBT dongles) moved
between BT and WiFi cells served by
MopogoBT dongles and CiscoAironet1100 APs,
respectively, changing to several access networks corresponding to different university
buildings. For the second experiment, instead,
we employed the IMS Bench SIPp, an IMS
traffic generator that conforms to the ETSI TS
186 008 IMS/NGN Performance Benchmark
specification. For IMS implementation, we
employed the OpenIMSCore that includes all
main IMS components and various facilities to
ease S-CSCF trigger creation as well as the
integration of new ASs [9]. The ASSC is implemented in Java, by exploiting the portable Java
application programming interface (API) for
Integrated Networks (JAIN) SIP implementation by the National Institute of Standards and
Technology. Our MNs exploit the open source
IMS Communicator — a pure Java IMS client
based on SUN Java Media Framework (JMF)
and JAIN SIP with IMS SIP extensions by
3GPP and IETF — that we originally extended
to support IHMAS optimizations [6, 10]. For
data buffering/streaming, our MP exploits
Asterisk, the widespread open source telephony
engine that we extended to support second-
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
level buffering [6]; the related performance
evaluation at the media delivery plane is out of
the scope of this article.
The results of the first set of experiments,
shown in Fig. 4, point out the effectiveness of
IHMAS handoff management techniques by
comparing their performance in the case of vertical handoffs from BT to WiFi. Reported results
are averaged over 1000 handoff cases, while provisioning an audio on demand service offering
G.711 (U-law) encoded flows (constant frame
rate = 50 frames/s) with buffer slots storing
G.711 RTP packets (216-byte payload/packet).
For each handoff management scenario, the figure shows session signaling duration (upper bar),
data handoff management delay (defined as the
delivery delay introduced by the applied handoff
technique, middle bar), and playout interruption
time (defined as the data duration of the media
gap perceived by final users, lower bar); the Ø
symbol represents negligible time intervals. Figure 4 focuses on a restricted time interval excluding VHP prediction: as known from our previous
experiments, our BT-to-WiFi handoff prediction
grants a prediction advance time between 5.4 s
and 6.8 s with sufficiently low error rates (under
9.0 percent) for most mobility conditions of
interest for multimedia streaming [6].
By focusing on session signaling, the reported
result is the sum of the duration of the four
main vertical handoff functions: data link connection, DHCP negotiation, REGISTER, and
INVITE signaling times. Data link connection
time strictly depends on handoff type, wireless
coverage, and client card characteristics [6]. For
basic handoff scenarios, it is very long (e.g., 670
ms), due to the employed reactive approach,
which activates the target WiFi network and
starts data link discovery only after connection
loss. In all other situations, instead, IHMAS
exploits VHP predictions to switch proactively to
WiFi as soon as a probable handoff is predicted,
thus eliminating discovery delays and dropping
data link connection time to 16 ms. DHCP negotiation time highly depends on DHCP
server/client implementation. We have extensively tested both Windows and Linux servers, and
decided to employ Linux dhcpd due to its
robustness. Moreover, while in the basic handoff
management case DHCP negotiation imposes
long delays (535 ms), our optimized techniques
accelerate the process by promptly activating
DHCP discovery as soon as the new interface is
active (delay down to 206 ms). Finally, to better
evaluate REGISTER and INVITE times when
compared with basic handoff, we do not include
the delay introduced by the distance between
CN and MN home networks because the ASSC
reconfigures the session directly at the MN
home. Under that assumption, the measured
REGISTER and INVITE signaling times are similar for all handoff management situations: 327
ms and 85 ms, respectively, as expected due to
the adoption of the standard IMS session signaling. The measurements exhibited a limited variance (under 7.5 percent in all cases).
The reported results clearly show that our
proactive approach can relevantly reduce session
signaling time. In addition, even most important,
it is demonstrated to be very effective for data
IEEE
BEMaGS
MN
Target P-/I-CSCFMN
MP
S-CSCFMN
ASSC
F
CN
Handoff
prediction
Datalink handoff
and DHCP req.
(1) REGISTER
(2) REGISTER
(3) 401 Unauth.
(4) 401 Unauth.
(5) REGISTER
(6) REGISTER
(7) 200 OK
(8) 200 OK
(9) INVITE
(10) INVITE
(11) INVITE (12) INVITE
Data flow over
old connection
Duplicates
elimination
(13) Soft
data handoff
trigger
Data flow
duplication and bind
(14) 200 OK
(16) 200 OK (15) 200 OK
(17) 200 OK
Data flow over
Old conn. target connection
loss
(18) Stop
data
duplication
Figure 3. Message flow for optimized soft data handoff management.
Basic handoff
management
Hard reactive
handoff
management
Data link connection
DHCP renegotiation
REGISTER signaling
INVITE signaling
Minimum service delay
Playout interruption
Hard proactive
handoff
management 0
Soft handoff
management 0
0
0
200
400
600
800 1000 1200 1400 1600 1800 2000
Time (ms)
Figure 4. Session signaling and data handoff completion time for BT to WiFi
handoff.
management delay and playout interruption. In
basic handoff, due to the serialization of handoff
operations, RTP connection rebinding can start
only after session signaling termination, thus
causing heavy data management delays (1812
ms). Our hard handoff management drops this
delay down to 220 ms, which corresponds to the
time required to send the INVITE message and
to finish RTP connection rebinding, started as
soon as MP receives the ASSC handoff trigger.
Moreover, if second-level buffering is used, it is
possible to mask the deriving loss of multimedia
data to final users at the expense of increased
packet delays (due to data buffering), thus avoiding any playout interruption. In addition, we
have performed preliminary experiments for
WiFi-to-3G vertical handoff, and the collected
results have demonstrated the same trends of
Fig. 4. The only main difference is a much longer
data link connection time (from 7.3 s without
optimizations to 4.2 s by properly decreasing the
wakeup period of 3G cards in idle state), while
the duration of other phases is approximately
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
119
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
the same. In particular, handoff prediction (in
this case measuring the possibility that WiFi signal strength at the MN drops below a given
threshold) permits very low latencies to be
achieved, as reported in Fig. 4 for the BT-toWiFi case. Regarding IHMAS soft handoff management, it supported lossless handoffs without
any additional frame delay cost in all the experiments. Finally, let us note that the performance
results in Fig. 4 largely outperform other stateof-the-art solutions in the literature (e.g., the
lowest MMD playout interruption time is 3666
ms). Such an improvement is possible because
IHMAS can eliminate data link connection and
MIP reconfiguration times, thanks to its signal-
220
200
10 cps
20 cps
30 cps
40 cps
50 cps
60 cps
70 cps
180
160
CPU %
140
S-CSCF
P-CSCF3
120
100
80
60
40
20
0
330,000 390,000 450,000 510,000 570,000 630,000 690,000 750000
ms
(a)
100,000
ms
10000
10 cps
20 cps
30 cps
40 cps
50 cps
60 cps
70 cps
INVITE w/o ASSC
INVITE with ASSC
1000
100
10
330,000 390,000 450,000 510,000 570,000 630,000 690,000 750,000
ms
(b)
220
200
180
CPU %
160
10 cps
20 cps
30 cps
40 cps
50 cps
60 cps
70 cps
S-CSCF
P-CSCF3
ASSC
140
120
100
80
60
40
20
0
330,000 390,000 450,000 510,000 570,000 630,000 690,000 750,000
(c)
ms
Figure 5. IHMAS session signaling scalability.
120
Communications
IEEE
A
BEMaGS
F
ing/media plane decoupling and application
layer approach [4].
Figure 5 reports the results of the second set
of experiments, aimed at quantitatively evaluating costs and scalability of ASSC interposition in
IHMAS in wide-scale deployment scenarios and,
more generally, of signaling overhead due to AS
interposition in IMS. By using the IMS Bench
SIPp and exploiting the concept of IMS session
phases supported in it, we defined a scenario
consisting of a first preparation phase, which
lasts 330 s and includes only IMS client registrations with a constant arrival rate of 15 calls per
second (cps), and a second evaluation phase,
which consists of a mix of 2.5 percent registrations, 2.5 percent deregistrations, 15 percent
reregistrations, 50 percent invitations, and 30
percent re-invitations. That mix mimics the usual
IMS traffic composition. The evaluation phase is
configured with 7 incremental steps of traffic,
going from 10 cps to 70 cps; each step (from 330
s to 750 s) lasts 60 s, and calls arrive according
to a Poisson distribution.
Delving into finer detail, we first stressed the
system without the ASSC within the path (Fig.
5a); then we repeated the experiments by interposing the ASSC and activating new IMS components to grant an infrastructure scalability
level comparable to the one obtained without
the ASSC (Fig. 5c). For the sake of simplicity,
here we focus on the most overloaded components, the S-CSCF, P-CSCF, and ASSC. Reported results show CPU utilization (from 0 to 200
percent, summing up the two CPUs) because, in
our experience, the CPU is the main bottleneck
in IMS infrastructures due to the costs of message parsing/forwarding. Other performance
indicators, such as memory and network load,
have proven less relevant to scalability (additional results are available at the IHMAS web site).
We also report the delay of the first INVITE signaling phase — the longest one (after the initial
unchanged REGISTER) due to initial filtering
criteria evaluation at S-CSCF and ASSC interposition, common to all IHMAS handoff strategies.
This delay is reported with and without ASSC
(dashed and solid lines in Fig. 5b, respectively, in
logarithmic scale). All results have exhibited limited variance (i.e., under 5 percent for 100 runs).
Without the ASSC, we have determined that
to stir one S-CSCF (solid line in Fig. 5a) to its
upper limit, it is necessary to deploy three PCSCFs (dashed line; we report only P-CSCF3
for sake of clarity). The S-CSCF starts collapsing
at 610 s (Fig. 5a), as also shown by the sudden
increment of the INVITE delay. At step 6, the
IMS Bench stops injecting new traffic due to the
number of received errors and S-CSCF saturation also provokes a message retransmission cascading effect on P-CSCFs, which collapse in the
following step (Figs. 5a and 5b). ASSC interposition produces a non-negligible message load
increase: each traversing SIP message provokes
the creation and injection of an additional SIP
message. However, to have performance results
comparable to the ones without the ASSC, it is
easy to activate an additional S-CSCF to split
and balance incoming load. With this simple
deployment variation, the S-CSCF begins to saturate at 530 s and IHMAS collapses at step 5,
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
with a cascading effect on the P-CSCF and ASSC
(Fig. 5c; the other S-CSCF exhibits a similar
behavior). The ASSC, instead, does not show
any scalability issues: before system collapse, its
CPU load is always below 80 percent.
In short, on one hand, the reported results
quantitatively show the non-negligible cost of AS
interposition in any IMS infrastructure in general. On the other hand, they demonstrate that
IHMAS scales well and can apply even to
deployment scenarios with high cps. Let us stress
that in realistic execution environments, IHMAS
optimization costs apply only to a subset of
incoming calls (in our experiments we considered all calls to rapidly stress our system); moreover, they are expected to be balanced by proper
pricing.
In addition, we did analogous performance
evaluations (both sets of experiments) for the
other vertical handoff direction (i.e., from WiFi
to BT): apart from longer BT data link connection delays, IHMAS enabled similar delay reductions in that case as well. For this reason, due to
space limitations, here we do not report those
results, which are available at the IHMAS Web
site, together with further implementation details
and performance evaluations.
CONCLUSIONS
The IHMAS research work demonstrates the
suitability of an application-level approach to
extend the standard IMS infrastructure in order
to improve its performance for session mobility
during handoffs. In particular, IHMAS flexibly
supports three data handoff management strategies that an ASSC can dynamically select and
configure depending on service requirements
and deployment conditions. That is possible by
preserving full compliance with the standard,
thus enabling the IHMAS deployment over
already installed IMS-conformant networks.
The encouraging results obtained with the
IHMAS prototype are motivating our current
research along two primary directions. On one
hand, we are executing objective measurement
of quality improvements/degradations during
vertical handoffs with/without multimedia adaptation at the MP. On the other hand, we are
using IHMAS to implement an IMS-based phone
meeting application for mobile users (support of
group communications).
IEEE
BEMaGS
F
REFERENCES
[1] G. Camarillo and M. A. García-Martín, The 3G IP Multimedia Subsystem (IMS), 2nd ed., Wiley, 2006.
[2] C. Kalmanek et al., “A Network-Based Architecture for
Seamless Mobility Services,” IEEE Commun. Mag., vol.
44, no. 6, June 2006, pp. 103–9.
[3] A. Udugama et al., “NetCAPE: Enabling Seamless IMS
Service Delivery across Heterogeneous Mobile Networks,” IEEE Commun. Mag., vol. 45, no. 7, July 2007,
pp. 84–91.
[4] A. Dutta et al., “Mobility Testbed for 3GPP2-Based Multimedia Domain Networks,” IEEE Commun. Mag., vol.
45, no. 7, July 2007, pp. 118–26.
[5] H. Schulzrinne, E. Wedlund, “Application-Layer Mobility
Using SIP,” ACM Mobile Comp. Commun. Rev., vol. 4,
no. 3, July 2000, pp. 47–57.
[6] P. Bellavista et al., “Context-Aware Handoff Middleware
for Transparent Service Continuity in Wireless Networks,” Pervasive Mobile Comp. J., vol. 3, no. 4, Aug.
2007, pp. 439–66.
[7] M. Toy, H. J. Stuttgen, and M. Ulema, “IP Multimedia
Systems in Infrastructure and Services — Part II,” IEEE
Commun. Mag., Special Issue on IP Multimedia Systems
and Services, vol. 45, no. 7, 2007, pp. 66–67.
[8] J. Rey et al., “RTP Retransmission Payload Format,” IETF
RFC 4588, July 2006.
[9] Open IMS Core Project; http://www.openimscore.org/
[10] IMSCommunicator Project; http://imscommunicator.
berlios.de/
______
The IHMAS research
work demonstrates
the suitability of an
application-level
approach to extend
the standard IMS
infrastructure in
order to improve its
performance for
session mobility
during handoffs.
BIOGRAPHIES
______________ graduated
PAOLO BELLAVISTA [SM] (paolo.bellavista@unibo.it)
from the University of Bologna, Italy, where he received a
Ph.D. degree in computer science engineering in 2001. He
is now an associate professor of computer engineering at
the University of Bologna. His research activities span from
mobile-agent-based middleware solutions and pervasive
wireless computing to location/context-aware services and
adaptive multimedia. He is a senior member of ACM. He is
an Editorial Board Member of IEEE Communications Magazine and IEEE Transactions on Services Computing.
ANTONIO CORRADI [M] (antonio.corradi@unibo.it)
______________ graduated
from the University of Bologna and received an M.S. in
electrical engineering from Cornell University, Ithaca, New
York. He is a full professor of computer engineering at the
University of Bologna. His research interests include distributed and parallel systems and solutions, middleware for
pervasive and heterogeneous computing, infrastructure
support for context-aware multimodal services, network
management, and mobile agent platforms. He is a member
of the ACM and the Italian Association for Computing
(AICA).
LUCA FOSCHINI [M] (luca.foschini@unibo.it)
____________ graduated from
the University of Bologna, where he received a Ph.D.
degree in computer science engineering in 2007. He is now
a research fellow of computer engineering at the University
of Bologna. His interests include distributed systems and
solutions for pervasive computing environments, system
and service management, context-aware session control
and adaptive mobile multimedia, and mobile-agent-based
middleware solutions. He is a member of AICA.
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
121
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
ACCEPTED FROM OPEN CALL
Reputation Estimation and Query in
Peer-to-Peer Networks
Xing Jin, Oracle USA
S.-H. Gary Chan, HKUST
ABSTRACT
Many peer-to-peer systems assume that peers
are cooperative to share and relay data. But in
the open environment of the Internet, there may
be uncooperative malicious peers. To detect
malicious peers or reward well behaved ones, a
reputation system is often used. In this article we
give an overview of P2P reputation systems and
investigate two fundamental issues in the design:
reputation estimation and query. We classify the
state-of-the-art approaches into several categories and study representative examples in each
category. We also qualitatively compare them
and outline open issues for future research.
INTRODUCTION
This work was supported,
in part, by the Hong Kong
Innovation and Technology Fund (ITS/013/08).
122
Communications
IEEE
In recent years peer-to-peer (P2P) systems have
seen enormous success and rich developments
over the Internet. Typical applications include
file sharing, streaming, Internet telephony, and
overlay routing. According to CacheLogic
Research, in 2006 P2P traffic accounted for over
72 percent of Internet traffic that year.
In P2P systems cooperative peers self-organize themselves into overlay networks and store
or relay data for each other. Many P2P systems
work on the assumption of truthful cooperation
among peers. However, in the open environment
of the Internet, some participating peers may
not cooperate as desired. They may be selfish
and unwilling to upload data to others, or they
may have abnormal actions such as frequent
rebooting that adversely affect their neighbors.
More seriously, some peers may launch attacks
to disrupt service or distribute viruses in the
overlay network. We call all these uncooperative, abnormal, or attacking behaviors malicious
actions and the associated peers malicious peers.
Malicious peers may seriously degrade the
performance of P2P networks. Liang et al. have
tracked several attacking behavior in practical
P2P file sharing systems [1]. They find that more
than 50 percent of copies of popular songs in
KaZaa are polluted, meaning that the content
downloaded from the network is different from
the downloader’s expectation (e.g., the content is
corrupt and cannot be played, or the content is a
different song from the search index metadata).
Their study also shows that both structured and
0163-6804/10/$25.00 © 2010 IEEE
unstructured P2P file sharing systems are highly
vulnerable if attackers insert massive bogus
records to poison search indexes.
To detect malicious peers or reward well
behaved ones, a reputation system is often used.
In a typical reputation system each peer is
assigned a reputation value according to its performance history. Differentiated services are
then provided to peers according to their reputation. While the basic idea is simple, a practical
system design is not easy. Generally, a P2P reputation system consists of three functional components [2]: collecting information on peer
behavior, scoring and ranking peers, and
responding based on peers’ scores. All these
components are nontrivial, especially given the
following consideration:
• Scalability: A large P2P network may have
hundreds of thousands of peers. For example, Skype has several million online users.
A reputation system should be highly scalable in terms of peer number.
• Adaption to peer dynamics: Peers may join
or leave at any time. If reputation information is maintained at peers, peer leaving
may lead to information loss. A robust reputation system should take peer dynamics
into account.
• Security: Malicious peers may endeavor to
break down the reputation system so that
they can conduct malicious actions without
being detected. For example, peers may
purposefully leave and rejoin the system
with a new identity in order to shed any
bad reputation [2]. Clearly, a good reputation system should be secure to resist these
adversarial behaviors.
In this article we study two fundamental issues
in P2P reputation systems.
Reputation estimation: An estimation method
describes how to generate peer reputation based
on others’ feedback. We classify existing estimation methods into three categories: social network,
probabilistic
estimation,
and
game-theoretic model. We select representative
examples from each category, and discuss their
advantages and limitations. As many estimation
methods rely on specific feedback collection
mechanisms, we also discuss feedback collection
mechanisms when necessary.
Reputation query: Reputation query in P2P
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
networks is not trivial. First, efficient data storage and retrieval is always a challenging issue in
P2P networks. Huge amounts of data require
distributed storage approaches. Then efficient
retrieval becomes nontrivial. Peer dynamics
bring more difficulty. Second, reputation data
are highly security-sensitive. The reputation of a
peer cannot be locally stored at the peer itself,
because a dishonest peer may misreport its reputation value in order to gain rewards or avoid
punishments. We also need to consider security
issues in reputation delivery. In this study we
survey the state-of-the-art approaches to reputation storage and retrieval in P2P networks. We
classify them into three categories. For each category, we discuss illustrative examples. We also
qualitatively compare them and outline possible
directions for future research.
There are many other important issues in
P2P reputation systems; for example, how to
prevent targeted and adversarial attacks? How
to interpret reputation? Interested readers may
refer to [2, 3] for a comprehensive overview of
P2P reputation issues.
The rest of the article is organized as follows.
In the next section we explore the reputation
estimation issue. We then discuss reputation
query techniques. We conclude in the final section.
REPUTATION ESTIMATION
There are mainly three reputation estimation
methods in current P2P networks. The first one
is the social network, where all feedback available in the network are aggregated to compute
peer reputation. The second one is probabilistic
estimation, which uses sampling of the globally
available feedback to compute peer reputation.
The third one is the game-theoretic model,
which assumes that peers have rational behavior
and uses game theory to build a reputation system. We elaborate on these methods below.
SOCIAL NETWORK
Approaches based on the social network can be
further divided into two categories: separated
reputation model and correlated reputation model.
In a separated reputation model only the direct
transaction partners of a peer (e.g., resource
provider/downloader or streaming neighbor) can
express their opinion on the reputation of the
peer. A practical example is the eBay reputation
system (although eBay is not a P2P network).
After each transaction at eBay, the buyer and
the seller rate each other with positive, negative,
or neutral feedback. The reputation is calculated
at a central server by assigning 1 point for each
positive feedback, 0 point for each neutral feedback, and –1 point for each negative feedback.
The reputation of a participant is computed as
the sum of its points over a certain period. Considering that peers may lie in their feedback,
Mekouar et al. propose to monitor suspicious
feedback [4]. That is, after each transaction
between a pair of peers, both peers are required
to generate feedback to describe the transaction.
If there is an obvious gap between the two pieces
of feedback, both are regarded as suspicious.
Later on, the more suspicious feedback a peer
generates, the smaller weight in reputation computing its feedback has. Similarly, in [5] a peer’s
reputation is computed as a weighted average of
feedback from direct witnesses of its performance. Xiong et al. develop a general reputation
model, which considers, for example, feedback
from peers, the trustworthiness factor of feedback sources, and the transaction context factor
for discriminating transaction importance [6].
Almost all separated reputation models can be
expressed by this model.
In a correlated reputation model the reputation of a peer is computed based on the opinion
of its direct transaction partners as well as thirdparty peers. In this model a peer, A, who wishes
to know the reputation of another peer, B, can
ask some peers (e.g., its neighbors) to provide
their opinion on B (although some of the peers
may not have conducted any transaction with
B). A then combines peer opinions to calculate
B’s reputation. We take EigenTrust as an example [7]. In EigenTrust, whenever a peer conducts a transaction with another peer, they keep
reputation values for each other. If there is no
direct transaction between two peers, they keep
a zero reputation value for each other. Peers
then iteratively update the reputation values.
Each time peer A wishes to update the reputation of peer B, A asks for B’s reputation from
all other peers in the system. A then computes a
weighted sum of these reputation values and
keeps the result as the new reputation of B. In
each iteration all peers conduct the above reputation update. The process continues until the
reputation values kept at different peers converge. Another example is the network information and control exchange (NICE) reputation
model [8]. Each peer holds the reputation of its
transaction partners according to the quality of
transactions. All peers further form a trust
graph based on reputation values. Later on, an
overlay path between two peers is selected as
the most trustworthy path between them in the
trust graph.
The correlated reputation model is more like
our real social network, where third-party peers
can express their opinion on a peer. But it costs
more to collect and aggregate third-party opinion. For example, EigenTrust takes a long time
to wait for reputation values to converge.
IEEE
BEMaGS
F
The correlated
reputation model is
more like our real
social network,
where third-party
peers can express
their opinion on a
peer. But it takes
more cost to collect
and aggregate
third-party opinion.
For example,
EigenTrust takes a
long time to wait
for reputation values
to converge.
PROBABILISTIC ESTIMATION
This approach uses sampling of the globally
available feedback to compute peer reputation.
It usually relies on some assumptions on peer
behavior. For instance, it may assume that a
peer is trustworthy with a certain but unknown
probability. And when sharing its own experience with others, a peer may lie with some,
again unknown, probability [9]. It then uses
probabilistic estimation techniques to estimate
all unknown parameters. Many estimation methods may be used. Despotovic et al. use maximum
likelihood estimation, which assumes that peers
do not collude [9]. Mui et al. use Bayesian estimation, which uses only direct interaction among
peers and does not use third-party opinion [10].
By using a small portion of the globally available feedback, the probabilistic model has lower
cost in feedback collection than the social net-
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
123
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
[55, 80) R
Step 1
Step 2
[95, 1000]
[20, 40)
Step 2
[0, 20)
T [80, 95)
[40, 55)
Step 3
B
Step 4
A
Proxies
Streaming peers
Figure 1. Process of submitting a report about a streaming peer A by its child B
for the first time in a proxy-based approach (from [5]). Step 1) B sends A's IP
address to R. Suppose that A's IP address is represented by a numerical value
88. Step 2) R searches in the binary tree to identify the proxy that manages 88
(T in this case). Step 3) T responds to B with its certificate. Step 4) After verifying the trustworthiness of T, B sends its report about A to T.
work approach. On the other hand, the social
network approach can use a complicated reputation model, and is robust to a wide range of
malicious actions. But the probabilistic model
can be applied to only simple reputation models
(due to the difficulty in probabilistic estimation)
and is effective against only a few kinds of malicious actions. The performance of the two models has been compared in [11]. It has been shown
that the probabilistic model performs better for
small malicious populations, while the social network approach is better when most peers are
malicious.
GAME-THEORETIC MODEL
Different from the above two approaches, the
game-theoretic model assumes that peers have
rational behavior and uses game theory to build
a reputation system. Rational behavior implies
that there is an underlying economic model in
which utilities are associated with various choices of peers and that peers act so as to maximize
their utilities. Fudenberg et al. present a gametheoretic framework to offer certain characterizations of the equilibria payoffs in the presence
of reputation effects [12]. But the work assumes
that a central trusted authority does feedback
aggregation, which may not be scalable to largescale P2P networks.
REPUTATION QUERY
In this section we discuss techniques for reputation query in P2P networks.
124
Communications
IEEE
A
BEMaGS
F
CENTRALIZED AND
PARTIALLY CENTRALIZED STRUCTURES
The simplest solution is to use a powerful server
to keep the reputation of all peers. For example,
eBay uses a central server to collect and keep all
users’ reputations. Feedback from users is sent
to and stored at the server. A query of a user’s
reputation is also sent to and answered by the
server. Similar approaches have been used in
[13].
A centralized approach is easy to implement
and deploy. Security of a central server is much
easier to achieve than that of distributed components in a distributed approach. Furthermore, in
a centralized approach, reputation management
is independent of peer joining and leaving, which
greatly simplifies reputation retrieval. However,
a centralized approach is not scalable to large
P2P networks. Also, the server forms a single
point of failure, making the system vulnerable.
To address the limitations of the centralized
approach, a partially centralized approach, which
uses a set of servers instead of a single server,
has been proposed. Mekouar et al. propose a
malicious detector algorithm (MDA) to detect
malicious peers in KaZaa-like systems [4].
KaZaa is a partially centralized P2P file sharing
system with a set of supernodes. Each ordinary
peer is attached to a unique supernode. MDA
assumes that supernodes are all trustworthy and
maintain reputation information for ordinary
peers. All evaluation results about a peer are
maintained at its attached supernode. Supernodes can then enforce differentiated service to
peers according to their reputation.
Note that supernodes in KaZaa are self-elected from ordinary peers and may not be fully
trustworthy. One approach uses predeployed
proxies instead of supernodes for reputation
maintenance [5]. In this approach each peer is
attached to a unique proxy according to its IP
address. Correspondingly, each proxy is responsible for a certain IP range, and proxies are
organized into a binary search tree based on the
IP ranges they maintain. Each peer periodically
generates reports about its streaming neighbors.
All reports about a peer are sent to its attached
proxy. A query about a peer’s reputation is also
forwarded to and answered by the peer’s
attached proxy. Figure 1 shows the report submission process in this approach [5]. Each circle
in the figure is a streaming peer, and each quadrangle is a deployed proxy. Numbers in a quadrangle indicate the IP range maintained by the
proxy (here numerical values are used to represent IP addresses). Suppose that streaming peer
B is streaming peer A’s child in the streaming
overlay, and B prepares to submit a report about
A’s performance. If B has not sent any report
about A before, B first sends A’s IP address to a
random proxy (which is R in the figure). R then
searches in the tree to identify the proxy whose
range covers A’s IP address (T in this case). T
then sends a response message to B as well as its
certificate of trustworthiness (issued by a trusted
certification authority). After B verifies the trustworthiness of T, it sends its report about A to T.
In the following, B will directly send reports
about A to T.
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Two important issues in partially centralized
approaches are efficient search and load balancing among multiple supernodes/proxies. First,
each peer should be attached to a unique supernode or proxy. In MD, this is done by KaZaa’s
built-in mechanism. If a P2P network does not
have such a built-in mechanism, this is not easy.
Suppose each proxy is responsible for a certain
range of peers. Given any peer in the system, we
need to quickly identify the proxy responsible for
it (e.g., for reputation update or query). If the
number of proxies is small, simple flooding can
be used for search. Otherwise, a more complicated overlay structure should be built among proxies (e.g., the binary search tree in [5]). Second,
loads for reputation management should be
evenly distributed among supernodes/proxies.
MDA does not consider this issue as it uses the
KaZaa built-in mechanism to attach peers to
supernodes. In [5] a dynamic load redistribution
method has been proposed to balance loads
among proxies.
Compared to centralized approaches, partially centralized approaches have significantly
improved system scalability. However, in order
to serve a large P2P network, a large number of
supernodes or proxies may be needed, which
leads to high implementation and maintenance
costs.
STRUCTURED OVERLAY
Another class of approaches uses distributed
hash table (DHT) to store and search for peer
reputation. In DHT each peer is assigned a
unique peer ID, and each object is hashed to a
key in the same space of peer IDs. The peer
with ID equal to the hashed key is responsible
for storing the location of the object (or the
object itself). With a hashed key of an object, a
query for the object is routed through peers in
DHT to the peer that is responsible for the
object. Each peer in DHT maintains a routing
table for routing messages.
We take PeerTrust as an example [6]. It
adopts P-Grid as the underlying DHT network.
It also uses a system-wide hash function Hash,
which maps one peer ID to another. Suppose
that peer p has an ID, ID(p). Whenever p has a
transaction with another peer, q, p generates a
report about q and sends it to the peer with ID
Hash(ID(q)) through DHT routing. The reputation of q is then stored and maintained at the
peer with ID Hash(ID(q)), which is called the
reputation manager of q. Queries of a peer’s
reputation are also forwarded to its reputation
manager through DHT routing. In this way, peer
reputation is distributedly stored in the system.
This approach has several advantages. First,
peer reputation is distributedly stored and computed at the reputation managers. There is no
need for a central server or supernodes. Second,
a peer’s reputation manager is determined by a
universal hash function, which cannot be selected by the peer itself. This reduces the possibility
of collusion between a peer and its reputation
manager.
However, this approach has some security
concerns. First, reputation managers may misbehave by providing false or random data when
answering a query. Majority voting has been
used to address this. That is, a DHT network
can be configured to have multiple replicas
responsible for the same key, or multiple hash
functions can be used to map each peer to multiple reputation managers [6]. When a peer
searches for the reputation of another peer, it
finds all the replicas responsible for the key and
uses a voting scheme to compute the final result.
However, voting cannot guarantee obtaining the
correct decision and does not completely address
the problem. As shown in [11], simple collusion
can seriously affect the result of voting. Second,
a reputation report or query is delivered between
its generator and the reputation manager by
DHT routing. A malicious peer in the delivery
path may modify, intercept, or discard the report
or query. PeerTrust has proposed to encrypt
messages in order to prevent data modification
during delivery [6]. But it cannot prevent data
discarding during routing. In summary, DHTbased approaches cannot guarantee secure reputation computing and delivery.
Furthermore, DHT has its own limitations.
Since peers are highly dynamic in P2P networks,
a reputation manager may unexpectedly leave
the system. Then the data maintained by it are
no longer available. In addition, load balancing
mechanisms that abide by DHT storage and
routing methods are complicated, especially in
dynamic networks. DHT also has its own security threats and vulnerabilities, and there are many
targeted attacks on its routing scheme, data
placement scheme, IP mapping scheme, and so
on.
IEEE
BEMaGS
F
As compared to centralized approaches,
partially centralized
approaches have
significantly
improved system
scalability. However,
in order to serve a
large P2P network,
a large number of
supernodes or
proxies may be
needed, which leads
to high implementation and maintenance costs.
UNSTRUCTURED OVERLAY
XREP uses a polling algorithm to choose reliable resource in Gnutella-like file sharing networks [14]. It consists of four operations:
resource searching, vote polling, vote evaluation
and resource downloading (Fig. 2). The first
operation is similar to searching in Gnutella. A
peer broadcasts to all its neighbors a Query
message containing the search keywords. When
a peer receives a Query message for which it
has a match, it responds with a QueryHit message, as shown in Fig. 2a. In the next operation,
upon receiving QueryHit messages, the query
initiator selects the best matching resource
among all possible choices. It then polls other
peers using an encrypted Poll message to
enquire about their opinion of the selected
resource or the resource provider. In XREP
each peer maintains information on its own
experience with the resource and other peers.
Upon receiving a Poll message, each peer
checks its experience data. If there is any information about the resource or the provider indicated by the Poll message, the peer sends its
vote to the polling peer with an encrypted
PollReply message, as shown in Fig. 2b.
In the third operation the polling peer collects a set of votes and evaluates the votes. It
first decrypts the votes and discards corrupt
ones. Then it analyzes voters’ IPs and detects
cliques of dummy or controlled votes. After that,
it randomly selects a set of votes and directly
contacts the voters with a TrustVote message.
Each contacted voter is required to send a
VoteReply message for vote confirmation. This
IEEE Communications Magazine • April 2010
Communications
A
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
125
A
BEMaGS
F
Communications
IEEE
B
A
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
ery
Qu yHit
er
Qu
Query
QueryHit
Q
Qu uer
er y
D
yH
it
(a)
B
B
C
A
ll
Po eply
llR
Po
Poll
PollReply
P
Po oll
llR
ep
ly
C
A
e
ot
stV eply
u
r
T
eR
ot
VTrustVote
F
B
C
A
C
Download
VoteReply
D
(b)
D
D
(c)
(d)
Figure 2. Operations in XREP: a) resource searching; b) vote polling; c) vote evaluation; d) resource downloading.
forces attackers to pay the cost of using real IPs
as false witnesses. After this checking process,
the polling peer can obtain the reputation of the
resource or provider. Based on the reputation
value, the polling peer can either download the
resource, as shown in Fig. 2d, or discard the
resource and repeat the voting process on another resource.
Approaches based on unstructured overlays
have similar limitations to DHT-based ones.
Messages may be intercepted or blocked during
transmission, and voting is vulnerable to collusion among peers. Therefore, no secure reputation computing or delivery can be guaranteed.
Furthermore, searching or voting on an unstructured overlay is based on flooding, which incurs
heavy traffic in the network. For example, in
XREP Poll messages are broadcast throughout
the network each time a peer needs to find out
the reputation of a resource or a provider.
COMPARISONS
We compare the above reputation query techniques in Table 1 and elaborate on the results
below.
A centralized approach requires a central
server for reputation storage, and a partially centralized approach relies on supernodes or predeployed proxies. On the contrary, approaches
based on structured and unstructured overlays
rely on peers to manage reputation and do not
require additional facilities. Specifically, in
DHT-based approaches a peer’s reputation is
maintained at its reputation manager, which is
computed by a universal hash function. In
approaches based on unstructured overlays,
peers often locally hold the reputation of their
transaction partners.
Based on different storage mechanisms, the
approaches have different reputation search
methods. In a centralized approach a reputation
query is directly sent to the server. In a partially
centralized approach a query is first sent to a
supernode, which forwards the query to the target supernode. In a DHT-based approach DHT
routing is used to route queries. In an approach
based on unstructured overlays, flooding is often
used, which may consume much network bandwidth.
Among these approaches, the centralized one
has the poorest scalability, while the DHT-based
one is the most scalable. The partially central-
126
Communications
IEEE
ized approach has better scalability than the centralized one, but still relies on predeployed proxies or supernodes and is not fully scalable. The
approach based on unstructured overlays does
not need any central component; however, it is
not as scalable as the DHT-based one because
of its high bandwidth consumption in reputation
search.
The centralized and partially centralized
approaches are robust to peer dynamics. In
these approaches reputation values are stored at
a server or supernodes, which are often highly
stable. In the DHT-based approach the leaving
of a reputation manager will lead to the loss of
data stored at it. Fortunately, DHT itself has
some mechanisms to keep high data availability
under peer churn. In the approach based on
unstructured overlays there is little protection
against data loss due to peer leaving. It may
encounter high data loss in the presence of peer
churn.
Regarding security, the centralized and partially centralized approaches are the most secure
if assuming the server and supernodes are fully
trustworthy. In these approaches reports or
queries are directly sent to the server or supernodes, and there are no third-party peers in
delivery paths. On the contrary, the approaches
based on structured or unstructured overlays
cannot guarantee secure reputation computing
or delivery. In these approaches a reputation
maintainer may be malicious and provide forged
data, and a delivery path may contain malicious
peers. Although there are many methods for
improving system security (e.g., encryption/
decryption or voting), none of them can guarantee 100 percent security.
CONCLUSION
In this article we investigate two key issues in
P2P reputation systems, reputation estimation
and query. We discuss representative examples
in the literature and compare them from multiple aspects. There are many other research
issues in P2P reputation systems, such as
anonymity. In many applications, users may only
be willing to participate if a certain amount of
anonymity is guaranteed. But most existing reputation systems have sacrificed anonymity in
order to provide secure underlying protocols,
where each peer holds a unique certificate, and
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
A
BEMaGS
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
F
Security
Approach
Deployment
requirement
Reputation
storage
Reputation
query
Scalability
Adaptation to
peer
dynamics
Centralized
A central
server
The server
Direct server
access
Low
Partially
centralized
A set of
supernodes
or proxies
Supernodes
or proxies
Search among
supernodes or
proxies
Medium
Structured
overlay
No
Peers
(computed
by a hash
function)
DHT search
Unstructured
overlay
No
Peers (e.g.,
transaction
partners)
Flooding
Trustworthiness of
reputation
maintainer
Message
modification
in
transmission
Message
discarding
in
transmission
High
Full
No (no overlay relay)
No (no overlay relay)
High
Full
No (no overlay relay)
No (no overlay relay)
High
Medium
No
guarantee
No
(addressed
by
encryption)
Possible
Medium
Low
No
guarantee
No
(addressed
by
encryption)
Possible
Table 1. Comparisons between various reputation query techniques.
peers use the certificates to authenticate each
other. Other interesting issues may include analyzing security threats and studying reward/punishment mechanisms.
[14] E. Damiani et al., “A Reputation-Based Approach for
Choosing Reliable Resources in Peer-to-Peer Networks,”
Proc. ACM CCS ‘02, Nov. 2002, pp. 207–16.
REFERENCES
X ING J IN [M] (xing.jin@oracle.com)
_____________ received his B.Eng.
degree in computer science and technology from
Tsinghua University, Beijing, China, in 2002, and his Ph.D.
degree in computer science and engineering from the
Hong Kong University of Science and Technology (HKUST),
Kowloon, in 2007. He is currently a member of technical
staff in the Systems Technology Group at Oracle, Redwood Shores, California. His research interests include
distributed information storage and retrieval, peer-to-peer
technologies, multimedia networking, and Internet topology inference. He is a member of Sigma Xi and the IEEE
Communications Society Multimedia Communications
Technical Committee.
[1] J. Liang et al., “Pollution in P2P File Sharing Systems,”
Proc. IEEE INFOCOM, 2005.
[2] S. Marti and H. Garcia-Molina, “Taxonomy of Trust: Categorizing P2P Reputation Systems,” Comp. Net., vol.
50, no. 4, Mar. 2006, pp. 472–84.
[3] S. Ruohomaa, L. Kutvonen, and E. Koutrouli, “Reputation Management Survey,” Proc. IEEE ARES ‘07, Apr.
2007, pp. 103–11.
[4] L. Mekouar, Y. Iraqi, and R. Boutaba, “Peer-to-Peer’s
Most Wanted: Malicious Peers,” Comp. Net., vol. 50,
no. 4, Mar. 2006, pp. 545–62.
[5] X. Jin, Q. Xia, and S.-H. G. Chan, “Building a Monitoring Overlay for Peer-to-Peer Streaming,” Proc. IEEE
GLOBECOM ‘06, Nov. 2006.
[6] L. Xiong and L. Liu, “PeerTrust: Supporting Reputationbased Trust for Peer-to-Peer Electronic Communities,”
IEEE Trans. Knowledge Data Eng., vol. 16, no. 7, July
2004, pp. 843–57.
[7] S. D. Kamvar, M. T. Schlosser, and H. Garcia-Molina,
“The EigenTrust Algorithm for Reputation Management
in P2P Networks,” Proc. WWW ‘03, 2003, pp. 640–51.
[8] R. Sherwood, S. Lee, and B. Bhattacharjee, “Cooperative
Peer Groups in NICE,” Comp. Net., vol. 50, no. 4, Mar.
2006, pp. 523–44.
[9] Z. Despotovic and K. Aberer, “Maximum Likelihood Estimation of Peers Performance in P2P Networks,” Proc.
P2PEcon ‘04, June 2004.
[10] L. Mui, M. Mohtashemi, and A. Halberstadt, “A Computational Model of Trust and Reputation,” Proc. IEEE
HICSS ‘02, Jan. 2002, pp. 2431–39.
[11] Z. Despotovic and K. Aberer, “P2P Reputation Management: Probabilistic Estimation vs. Social Networks,”
Comp. Net., vol. 50, no. 4, Mar. 2006, pp. 485–500.
[12] D. Fudenberg and D. Levine, “Reputation and Equilibrium Selection in Games with a Patient Player,” Econometrica, vol. 57, no. 4, 1989, pp. 759–78.
[13] S. Jun, M. Ahamad, and J. Xu, “Robust Information
Dissemination in Uncooperative Environments,” Proc.
IEEE ICDCS ‘05, June 2005, pp. 293–302.
BIOGRAPHIES
S.-H. G ARY C HAN [M] (gchan@cse.ust.hk)
___________ received his
B.S.E. degree (Highest Honor) in electrical engineering
from Princeton University, New Jersey, in 1993, with certificates in applied and computational mathematics,
engineering physics, and engineering and management
systems, and his M.S.E. and Ph.D. degrees in electrical
engineering from Stanford University, California, in 1994
and 1999, respectively, with a minor in business administration. He is currently an associate professor with the
Department of Computer Science and Engineering,
HKUST, and an adjunct researcher with Microsoft
Research Asia, Beijing. His research interests include multimedia networking, peer-to-peer technologies and
streaming, and wireless communication networks. He is a
member of Tau Beta Pi, Sigma Xi, and Phi Beta Kappa.
He served as a Vice-Chair of the IEEE Communications
Society Multimedia Communications Technical Committee from 2003 to 2006. He was a Guest Editor for IEEE
Communications Magazine, Special Issue on Peer-to-Peer
Multimedia Streaming (2007), and Springer Multimedia
Tools and Applications, Special Issue on Advances in
Consumer Communications and Networking (2007). He
was Co-Chair of the Multimedia Symposium for IEEE ICC
2007. He was Co-Chair of the workshop on Advances in
Peer-to-Peer Multimedia Streaming at ACM Multimedia
2005, and the Multimedia Symposia for IEEE GLOBECOM
2006 and IEEE ICC 2005.
IEEE Communications Magazine • April 2010
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
127
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
ADVERTISERS’ INDEX
Company
Page
ADVERTISING SALES OFFICES
Agilent Technologies ................................................................3
Asilomar Conference................................................................17
Cisco ...........................................................................................Cover 2
GL Communications ................................................................8
ICC 2010 ....................................................................................1
International Microwave Symposium .....................................41
Closing date for space reservation:
1st of the month prior to issue date
NATIONAL SALES OFFICE
Eric L. Levine
Advertising Sales Manager
IEEE Communications Magazine
3 Park Avenue, 17th Floor
New York, NY 10016
Tel: 212-705-8920
Fax: 212-705-8999
e-mail: _____________
e.levine@comsoc.org
SOUTHERN CALIFORNIA
Patrick Jagendorf
7202 S. Marina Pacifica Drive
Long Beach, CA 90803
Tel: 562-795-9134
Fax: 562-598-8242
patrick@roymcdonald.com
________________
John Wiley & Sons, Ltd............................................................Cover 3
Korea Telecom ..........................................................................13
Samsung .....................................................................................Cover 4
NORTHERN CALIFORNIA
George Roman
4779 Luna Ridge Court
Las Vegas, NV 89129
Tel: 702-515-7247
Fax: 702-515-7248
Cell: 702-280-1158
george@roymcdonald.com
_________________
SK Telecom................................................................................5
Smart Grid Conference 2010 ...................................................23
SOUTHEAST
Scott Rickles
560 Jacaranda Court
Alpharetta, GA 30022
Tel: 770-664-4567
Fax: 770-740-1399
___________
srickles@aol.com
EUROPE
Rachel DiSanto
Huson International Media
Cambridge House, Gogmore Lane
Chertsey, Surrey, KT16 9AP
ENGLAND
Tel: +44 1428608150
Fax: +44 1 1932564998
Email: rachel.disanto@husonmedia.com
____________________
128
Communications
IEEE
IEEE Communications Magazine • April 2010
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
New and Bestselling from
Wiley-IEEE Press
A Guide to the Wireless
Engineering Body of Knowledge
(WEBOK)
The ComSoc Guide to Next
Generation Optical Transport
IEEE COMMUNICATIONS SOCIETY
The ultimate reference book for professionals
in the wireless industry and study guide for the
WCET. The information presented in this book
reflects the evolution of wireless technologies,
their impact on the profession, and the industry’s
commonly accepted best practices.
HUUB VAN HELVOORT
Provides a unique overview of SDH and OTN for
engineers who are new to the field, as well as
manufacturers, network operators, and graduate
students who need a basic understanding of the
topics.
978-0-470-43366-9 • April 2009 • Pbk • 272pp
$69.95
SDH/SONET/OTN
978-0-470-22610-0 • October 2009 • Pbk • 211pp
$63.50
ComSoc Guides to Communications
Technologies
Handbook on Array Processing and
Sensor Networks
Wireless Sensor Networks
SIMON HAYKIN and K. J. RAY LIU
Provides readers with a collection of tutorial articles
contributed by world-renowned experts on recent
advancements and the state of the art in array processing
and sensor networks.
JUN ZHENG and ABBAS JAMALIPOUR
This book provides a comprehensive and systematic
introduction to the fundamental concepts, major
challenges, and effective solutions in wireless sensor
networking (WSN).
978-0-470-37176-3 • January 2010 • Hbk • 904pp
$185.00
Adaptive and Learning Systems for Signal Processing, Communications
and Control Series
978-0-470-16763-2 • October 2009 • Hbk • 489pp
$94.95
A Networking Perspective
Ground-Based Wireless Positioning
Advances in Multiuser Detection
MICHAEL L. HONIG
During the past decade, the design and development of
current and emerging wireless systems have motivated
many important advances in multiuser detection. This
book provides a comprehensive overview of crucial recent
developments
978-0-470-47381-8 • September 2009 • Hbk • 493pp
$125.00
Wiley Series in Telecommunications and Signal
Processing
Next Generation Solutions
TULAY ADALI and SIMON HAYKIN
Recent developments have made it clear that significant
performance gains can be achieved beyond those using
standard adaptive filtering approaches. This book presents
the next generation of algorithms that will produce these
desired results.
978-0-470-19517-8 • April 2010 • Hbk • 424pp
$120.00
Adaptive and Learning Systems for Signal Processing, Communications
and Control Series
IEEE
Wireless Positioning
Kegen Yu Ian Sharp Y. Jay Guo
978-0-470-74704-9 • June 2009 • Hbk • 450pp
$120.00
Near-Capacity Multi-Functional
MIMO Systems
LAJOS HANZO, OSAMAH ALAMRI, MOHAMMED
EL-HAJJAR and NAN WU
Providing an all-encompassing self-contained treatment
of Near-Capacity Multi-Functional MIMO Systems it
gives a detailed examination of wireless landscape,
including the fields of channel coding, spacetime coding and turbo
detection techniques.
978-0-470-77965-1 • May 2009 • Hbk • 738pp
$200.00
ORDER WILEY ONLINE…
For telephone orders:
Call 1-800-225-5945 (1-800-US-WILEY)
Fax: (212) 850-8888
E-mail:_________
custserv@wiley.com
• Visit the Wiley Electrical Engineering homepage: www.wiley.com/electrical
• Wiley Communications Technology Website: www.wiley.com/go/commstech/
• Amazon Wiley Communications Technology bookstore: www.amazon.com/professional/
Please note all prices correct at time of going to press but subject to change.
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
14140
HOW TO ORDER
Communications
Ground-Based
Sphere-Packing, Iterative Detection and
Cooperation
Adaptive Signal Processing
Wiley books are available through your Bookseller.
Alternatively send your order to:
John Wiley & Sons Inc., 111 River Street, Hoboken,
NJ 07030, USA
KEGEN YU, IAN SHARP and Y. JAY GUO
Provides an in-depth treatment of non-GPS based
wireless positioning techniques, with a balance between
theory and engineering practice. The book presents the
architecture, design and testing of a variety of wireless
positioning systems based on the time-of-arrival, signal
strength, and angle-of-arrival measurements.
BEMaGS
F
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F
__________________________
______
Communications
IEEE
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
A
BEMaGS
F