ITCE 720A Autonomic Wireless
Networking
Fall 2009
Prof. Chansu Yu
chansuyu@gmail.com
c.yu91@csuohio.edu
1
Who am I
Industry Experience at GoldStar/LG
(1984-1997)
Handheld PC at LG Inc.
Microsoft’s Windows CE
 Hitachi’s SH3 Microprocessor

Design Issues

Interrupt latencies
 Power modes & Energy saving
 Cache performance (Accessible cache)
 ARM Support (Virtual cache)
2
Who am I
PC BIOS (Basic Input/Output Systems) Project at GoldStar
Inc.
 Clean room process
 Specification
writing team : aware of BIOS code
 Bios writing team : virgin engineers
I
was in the Specification team to work on POST
(Power-on-self-test) & HDD (hard disk drive)
3
Who am I
Academic experience

ICU (1998-2001)
 Cleveland State University (2001-Current), USA
Projects in academia

WebCam with Blue Cord Technology Co.
 Remote monitor with ETRI
 Energy-aware mobile networking
 TDMA-based sensor networks
Current projects





Performability in Mobile Wireless Networks (NSF)
Exploring Data Access in Internet-based Wireless Mobile Networks (NSF)
Low-power Wireless Networking in Software Radio Systems (Fenn)
Improving Work Zone Safety using Sensor Networks (CSU)
Seamless Connectivity and High Fidelity Communications in Multihop Wireless
Mesh Networks (NSF, pending)
4
Cleveland, OHIO
1796: Established in 1796 by Cleveland
1930: City population is 1M and the 5th
2008: City population is 0.5M
1951: Disc jockey Alan Freed at radio
station WJW began playing a certain
type of music for a multi-racial
audience. Freed coined the phrease
"rock and roll" to describe the rollicking R&B music. He
organized the first rock and roll concert called "The Moondog
Coronation Ball“
Severance Hall (1931, Cleveland Orchestra)
Cleveland Clinic (1921, Ranking 3 after Johns Hopkins and Mayo)
5
LeBron James (24), Cleveland Cavaliers
The No. 1 overall pick in the 2003 NBA
draft ($20M/year contract @ 2008, one
of the highest-paid athlete in US)
1983: Gordon Gund purchased @ $20M
2005: Gordon Gund sold @ $375M
Value @ 2007: $455M ($202M @ 2002)
32-of-41 regular season games were sold
out (50% in 2002)
17 games televised in China (0 in 2002)
Cleveland Cavaliers won the first Eastern
Conference Championship in 2007
6
Shin-Soo Choo (27), Cleveland Indians
7
Introduction
8
Evolution of Computing
Single user systems
Batch processing
Time-sharing
Networked computing
Mobile computing & Ubiquitous computing
Autonomic computing
(mobile devices + wireless communication + autonomic networking)
9
Mobile Computing
Goal

Access information anywhere, anytime
Aliases

Nomadic computing, wireless computing, ubiquitous computing, wearable
computing
Entire new class of applications

New massive markets
 Personal computing + consumer electronics + wireless communication
 Collaborative computing, vehicle dispatching, point of sale, mail enabled
applications, filtered information provision, …
10
1. Mobile Computers
11
Information at your Fingertips
Handheld or Pocket PC, PDA

PC players - easy interface on a computer
: Apple Newton, HP Palmtop, Microsoft Windows CE

Electronics players - computing on home appliance
: Sharp Zaurus, Philips Velo, Psion Series, 3Com (USR) PalmPilot
Starts from and targets PC metaphor (users)





Shrunken PC user interface : ex) stylus for mouse
Simple PIM applications : Personal Information Mgmt
Strong connectivity to Windows PC via IR or Serial
PCMCIA, USB, IrDA, Mini-card, Flash ROM
Stand-alone device but can add networks
12
Mobility Issues
Bandwidth restrictions and variability
Location-aware network operation

User may wake up in a new environment
 Dynamic replication of data
Querying wireless data & location-based responses
Bursty network activity during connections & handling disconnections
Disconnection

OS and File System Issues - allow for disconnected operation
 Database System Issues - when disconnected, based on local data
13
Portability Issues
Battery power restrictions
Risks to data

Physical damage, loss, theft
 unauthorized access
 encrypt data stored on mobiles
 backup critical data to fixed (reliable) hosts
Small user interface

Small displays due to battery power and aspect ratio constraints
 Cannot open too many windows
 Difficult to click on miniature icons
 Input Graffiti, (Dictionary-based) Expectation
Gesture or handwriting recognition with Stylus Pen
Voice matching or voice recognition
14
Portability Issues : Power Management
Key Ideas

Subsystems may have small duty factor
 Power down individual components when they are idle
Approach

Go to reduced mode after idle for a few time
 Predictive approach : use history to predict
 But,


Cost of restarting : latency and power
Alternatives


Pre-wakeup
Dynamic Voltage Scaling
– Intel SpeedStep
– Transmeta Crusoe
CPU
Speed
CPU
Speed
15
Same area = same work
But energy is saved due to low voltage
Portability Issues : Power Management
802.11 (WaveLAN-II)
Hardware State
Awake
Bluetooth (Nokia)
Mode of Operation Mode of Operation Hardware State
Transmit
(300mA)
Receive
Active
(250mA)
Idle(Listen)
(230mA)
Power Save
Active
(40-60mA)
Connection
Sniff
Hold
Doze
Park
Sleep
(9mA)
Standby
(0.55mA)
Standby
768 Kbps, 10-100 meters
2 Mbps, 250 meters
16
2. Wireless Communication
Main research challenges due to mobility

variable communication conditions
 energy limitations
Effects on different layers of OSI hierarchy

mobile communication : physical/MAC layer
 mobile computing : data link/network/transport layer
Research Issues

Mobile Networking - Network Layer





Mobile IP
Location Management
Multicasting
Ad-hoc networking
Mobile Networking - Transport Layer
17
Mobile Networking
How the network/transport layer protocols are affected in mobile and
wireless environment
Mobility Management - Network Layer

Mobile unit's physical location is no longer determines its network address
- does not know



where a given user is
how to route messages
Approaches


Internet community: mobile IP (extends IP, connectionless)
Cellular communication community: location management (connection-oriented)
18
Location (Mobility) Management
Mobility management

Find an adequate tradeoff between searching and informing


Searching by the system
Informing by the mobile hosts
– when MHs receives messages frequently
– when MHs does not move between cells often
Multicasting

Possible to receive no / multiple identical messages
 Multicasting is a challenge



how to guarantee “exactly once” or “at least once” delivery in an efficient manner
MCAST protocol is proposed
how to maintain a “group view” - the set of MSSs for multicasting
19
Mobile Ad-hoc Networking
Ultimate challenge for mobile networking
Mobile terminals can form networks without participation of the fixed
infrastructure, arise in rapid-deployment situations

emergency service at a disaster site
 military operations in a remote area
 business meetings held in venues without network infra
 sensor networks
Highly dynamic

frequent change of routing table
 a given terminal can serve as a router now but no longer be a short time
later
20
Mobile Ad-hoc Networking
21
3. Embedded Systems
Hardware

Embedded processors & embedded peripherals
 Interfacing




Interacts with environments
Sensing and controlling externals
Real-time constrains
I/O : System bus, I2C, Parallel, Serial, IR, RF, PCMCIA
Software

Embedded OS



Palm OS, Windows CE, Embedded Linux, …
Cross development & Emulation
Workload characterization
22
David Tennenhouse, “Proactive computing,” Communications of the ACM, Vol. 43, No. 5, pp. 43-50, May 2000
23
Karen W. Markus and Kaigham J. Gabriel, “MEMS: The Systems Function Revolution,” IEEE Computer, pp.25-31, Oct. 1999.
24
25
Market Fragmentation (RTOS)
In House
ISI - 6.0%
WRS - 4.6%
Microtec - 3.7%
Microware - 3.6%
QNX - 2.6%
Lynx - 2.6%
Concurrent - 2.5%
All other commercial
RTOS - 24.4%
26
 Severely Fragmented
 In-house trend indicative
that “One Size” does not
fit all
 Market trends indicates
value in tool chain and
applications, NOT in
kernel
Workload Characterization
Jakob Engblom Studied Embedded Programs

13 applications, 337 kloc
 Various industrial applications:

Telecomm, Vehicles, Consumer Products, …

Embedded, partially real-time programs
 Medium-capacity 8- and 16-bit CPUs:


Z80, 68HC11, C166, MELPS7000, H8, …
Medium-to-large European companies
27
Variables: Integers
 SpecInt95 integers
 Embedded integers
100%
100%
90%
90%
80%
80%
70%
70%
unsigned
signed
60%
50%
50%
40%
40%
30%
30%
20%
20%
10%
10%
0%
0%
char
short
unsigned
signed
60%
long
char
28
short
long
Parameters and Return Values
 SpecInt95
param &
return
64%
 Embedded
void-void
35%
return value
15%
return value
4%
void-void
4%
param &
return
64%
parameters
28%
parameters
25%
29
4. Ubiquitous Computing
Whenever people learn something sufficiently well,
they cease to be aware of it
Location-based services, shared meeting applications
<-> Virtual reality : make a world inside the computer
Hundreds of computers in a room “share situations”
Examples

Active Badge - door control, phone call forwarding, terminal preference
 Tab - instant votes, library map
30
G. W. Fitzmaurice, “Situated Information Spaces and Spatially Aware Palmtop Computers,” Communications of the ACM,
Vol. 36, No. 7, pp. 39-49, Jul. 1993 (this issue contains many other related articles including M. Weiser’s).
31
32
Active Badges
Purpose

locating individuals within a building by determining the location of their Active
Badge (and thus, telephone calls can be routed)
Method

This small device worn by personnel transmits a unique infra-red signal every 10
seconds (for 0.1 sec to reduce collision probability)
 Each office within a building is equipped with one or more networked sensors
which detect these transmissions
 The location of the badge (and hence its wearer) can thus be determined on the
basis of information provided by these sensors
33
M. Spreitzer and M. Theimer, “Providing Location Information in a Ubiquitous Computing Environment,” Mobile
Computing, Edited by T. Imielinski and H. F. Korth, Chapter 15, Kluwer Academic Pub., 1996.
34
Hiroaki Koshima & Joseph Hoshen, “Personal locator services emerge,” IEEE Spectrum, pp. 41-48, Feb. 2000.
Geolocation system architecture links the user of a
locator with a subscriber to the service by way of
location service providers, a location center, and a
wireless network. The arrows in the diagram represent
the flow of data for a subscriber [lower left] seeking a
user.
35
5. Autonomic networking
integration of heterogeneous fixed and
mobile networks with varying
transmission characteristics
regional
vertical
handover
metropolitan area
campus-based
in-house
36
horizontal
handover
Wireless systems: overview of the development
cellular phones
satellites
1982:
Inmarsat-A
1983:
AMPS
1984:
CT1
1986:
NMT 900
1987:
CT1+
1988:
Inmarsat-C
1991:
CDMA
1991:
D-AMPS
1989:
CT 2
1992:
Inmarsat-B
Inmarsat-M
1993:
PDC
1994:
DCS 1800
analogue
wireless LAN
1980:
CT0
1981:
NMT 450
1992:
GSM
cordless
phones
1991:
DECT
1998:
Iridium
2000:
GPRS
1997:
IEEE 802.11
1999:
802.11b, Bluetooth
2000:
IEEE 802.11a
2001:
IMT-2000
digital
200?:
Fourth Generation
(Internet based)
37
199x:
proprietary
IEEE 802.11 Protocols

IEEE 802.11a


IEEE 802.11b


PHY Standard: 3 channels : OFDM and PBCC
IEEE 802.11h


Inter-Access Point Protocol
IEEE 802.11g


MAC Standard : QoS support
IEEE 802.11f


MAC Standard : operate in variable power levels
IEEE 802.11e


PHY Standard : 3 channels : 11 Mbps
IEEE 802.11d


PHY Standard : 8 channels : 54 Mbps
Supplementary MAC Standard: TPC and DFS
IEEE 802.11i

Supplementary MAC Standard: Alternative WEP
38
Typical application: road traffic
UMTS, WLAN,
DAB, GSM,
cdma2000, TETRA, ...
Personal Travel Assistant,
DAB, PDA, laptop,
GSM, UMTS, WLAN,
Bluetooth, ...
39
Autonomic Wireless Networking
Diverse systems and solutions motivate us to consider self-* networks
Applications and services are not ported onto a preexisting network, but
the network itself grows out of the applications and the services that
end users want
Service-driven, situated, self-controlled, self-organized, technologyindependent, and scalable
Network is not just a “data transport engine” but more like an “information
transport/processing engine”
40
Software-defined and Cognitive Radios
Cognitive radio is “an intelligent wireless communication system that is
aware of its environment … adapts to statistical variations in the input
stimuli, with two primary objectives in mind – highly reliable
communication and efficient utilization of the radio spectrum.” (Simon
Haykin, IEEE JSAC, 2005)
Three fundamental cognitive tasks are
radio-scene analysis,
channel-state estimation and predictive modeling, and
transmit power control and dynamic spectrum management
Technical foundation of cognitive radio is software-defined radio platform
GNU Radio and USRP
SORA (MSR China)
CalRadio (UC)
Etc.
41
USRP & GNU Radio
USRP Hardware - Block Diagram
Fig. source - http://www.nd.edu/~jnl/sdr/docs/tutorials/4.pdf
Architecture
Sender
User-defined
Code
USB
PC
DAC
FPGA
USRP (mother board)
RF
Front end
USRP
(daughter board)
Receiver
User-defined
Code
USB
FPGA
ADC
43
RF
Front end
ACM MobiCom’07
44
ACM SigComm’07
45
IEEE Workshop ..,
2006
46
BBN Technical Memo, 2006
Revised and published in
IEEE MILCOM, 2007
47