PERFORMANCE EVALUATION OF
INTEGRATED WiMAX AND WLAN
NETWORKS FOR VOICE OVER IP
APPLICATION
D ISSERTATION S UBMITTED
THE
M.TECH
IN
IN PARTIAL FULFILLMENT FOR
D EGREE
OF
D ISTRIBUTED AND M OBILE C OMPUTING
IN
JADAVPUR UNIVERSITY
BY
ANINDITA KUNDU
Under the guidance of
P ROF . I TI S AHA M ISRA
DEPARTMENT OF ELECTRONICS &
TELECOMMUNICATION ENGINEERING
JADAVPUR UNIVERSITY
KOLKATA-32
MAY 2010
FACULTY OF ENGINEERING AND TECHNOLOGY
JADAVPUR UNIVERSITY
CERTIFICATE
I hereby recommend that the thesis prepared under my supervision by
Anindita Kundu (Reg No.: 105355 of 2008-09, Class Roll No: 000812302006,
Exam Roll No: M4DMC10-05) entitled “Performance Evaluation of Integrated
WiMAX and WLAN networks for Voice over IP application” be accepted in
partial fulfillment of the requirements for the degree of Master of Technology in
Distributed & Mobile Computing of Jadavpur University.
-------------------------------------(Prof. Iti Saha Misra)
Supervisor
Dept. of Electronics and
Telecommunication Engineering
Jadavpur University
Kolkata - 700032
COUNTERSIGNED
------------------------------------------------------------------(Prof. Nandini Mukherjee)
Director, School of Mobile
Computing and Communication
And
Professor, Department of
Computer Science and Engineering
Jadavpur University
Kolkata - 700032
---------------------------------------------
(Prof. Niladri Chakraborty)
Dean, Faculty of Engineering
and Technology
Jadavpur University
Kolkata - 700032
ii
FACULTY OF ENGINEERING AND TECHNOLOGY
SCHOOL OF
DISTRIBUTED & MOBILE COMPUTING
JADAVPUR UNIVERSITY
Certificate of Approval *
The foregoing thesis is hereby approved as a creditable study of an engineering
subject, carried out and presented in a manner satisfactory to warrant its acceptance
as a prerequisite to the degree for which it has been submitted. It is understood that
by this approval the undersigned do not necessarily endorse or approve any
statement made, opinion expressed or conclusion drawn therein but approve the
thesis only for the purpose for which it has been submitted.
Final examination evaluation of
Thesis of
Signature of the Examiner
Signature of the Supervisor
*Only in case the thesis is approved
iii
Acknowledgement
First and foremost, I would like to express my sincere gratitude to my supervisor,
Professor Iti Saha Misra for her valuable guidance and insightful suggestions throughout my
thesis work. I am very much indebted to her for giving me the opportunity to pursue my M.Tech
thesis work under her guidance. Without her precious knowledge in the field of wireless
communication and networks and exceptional encouragement my project work would not have
taken a meaningful shape. She has always encouraged me to get my work published in various
conferences. I will be benefited throughout my life from the knowledge, rightful guidance and
unparallel professionalism that I acquired here.
I have learnt quite a lot about the network simulator OPNET without which the
development of my M.Tech project would have very tough. I deeply acknowledge the support
from DST, Govt. of India for this work in the form of FIST 2007 Project on, “Broadband
Wireless Communications” in the Dept. of ETCE, Jadavpur University and thereby I also
extend my thanks and appreciation towards Professor Salil K. Sanyal and Professor Iti Saha
Misra of Dept. of ETCE, Jadavpur University, for their contributions in making OPNET
available in the laboratory through the DST project.
It is beyond my literal and material means to express my heartfelt thanks and deep
gratitude to Professor Salil K. Sanyal for his insight, guidance and advise which was extremely
important in this project in giving it the proper direction and necessary perspective.
I was extraordinarily fortunate to have Professor Nandini Mukherjee as the director of
the School of Mobile Computing and Communication and thank her for all the support she has
provided during the entire duration of the project.
It is my pleasure to acknowledge my friend Suman Bhunia who was also pursuing his
master's project in the OPNET laboratory, my junior students Basabdatta Palit, Atri
Mukhopadhyay and Tamal Chakraborty for their extraordinary support and co-operation. I
also thank all my good friends Sayantani Saha, Kamalesh Karmakar, Ishita Bhakta, Nilesh
Das, Shubhankar Sarkar, Prasun Chowdhury and Budhaditya Bhattacharyya whom I have met
here during the course and with I have shared unforgettable cheerful memories.
Finally, I thank my parents who have always supported me with deep love and
sacrifice throughout my life. I pray for their happiness and good health and I dedicate my work
to them in the most sincere way I can think of.
Place: Kolkata
Date:
(Anindita Kundu)
School of Mobile Computing and
Communication
Regn No.: 105355 of 2008-09
Class Roll No.: 000812302006
Exam Roll No.: M4DMC10-05
Jadavpur University
Kolkata - 700032
iv
Abstract
With wireless communication becoming an integral part of human life, the
improvement of the performance of any wireless network has become a topic of
keen interest of the researchers. The path of propagation being wireless, the
performance of the network is affected consequently by the topology and the
environmental conditions of the area where the network is deployed. Hence, a
study of the performance of the widely researched WiMAX network is performed
under varying terrain and environmental conditions for various propagation models
in this thesis. Also with 4G mobile networks coming up, the performance of the
integrated WLAN-WiMAX network is also evaluated and compared with the
existing WLAN and WiMAX technologies. Voice over IP is considered as the
application as it is expected to be a low cost and thereby, a popular communication
system in the next generation communication networks. The digitization of the
analog voice signal before transmission is done by the voice codecs. Hence, a
study of the performance of the networks and VoIP is also conducted here with
different codecs for the three technologies. It is observed that the network
performance is best for free space path loss model as it considers the
communication path to be free from obstacles. Also the WLAN-WiMAX
integrated network is observed to perform best among the three technologies with
respect to network capacity and network performance.
v
List of Tables
1. TABLE 1.1. Bandwidth Requirement of Some Common Codecs
2. TABLE 2.1. Numerical Values Considered for the Parameters
3. TABLE 2.2. Network Configuration Details
4. TABLE 2.3. Attributes of the Network Components
5. TABLE 3.1. Basic Characteristics of various IEEE 802.16 standards.
6. TABLE 4.1. Characteristic feature of Wireless LAN
7. TABLE 5.1. WLAN and WiMAX Comparison
8. TABLE 6.1. Tunnel Interface Listing
9. TABLE 6.2. Interface Information
vi
List of Figures
1. FIGURE 1.1. Wi-Fi integration using WiMAX
2. FIGURE 1.2. Packet Creation based on voice activity.
3. FIGURE 2.1. Network Model for WiMAX.
4. FIGURE 2.2. Path loss due to various pathloss models in decibel
5. FIGURE 2.3. WiMAX Network Throughput in Bits per second
6. FIGURE 2.4. Packet end to end delay in seconds
7. FIGURE 3.1. Reference Model of IEEE 802.16
8. FIGURE 3.2. Schematic Model of WiMAX PHY Layer
9. FIGURE 3.3. WiMAX Network Architecture
10. FIGURE 3.4. Network Model for WiMAX in OPNET
11. FIGURE 3.5. Base Station Parameters
12. FIGURE 3.6. Subscriber Station Parameters
13. FIGURE 3.7. Average voice jitter without Silence Suppression
14. FIGURE 3.8. Average voice jitter using Silence Suppression
15. FIGURE 3.9. Average packet end to end delay without Silence Suppression
16. FIGURE 3.10. Average packet end to end delay using Silence Suppression
17. FIGURE 3.11. Average MOS without Silence Suppression
18. FIGURE 3.12. Average MOS with Silence Suppression
19. FIGURE 3.13. Network Model for WiMAX
20. FIGURE 3.14. Mean Jitter without Silence Suppression
21. FIGURE 3.15 Mean Packet end to end delay Without Silence Suppression
22. FIGURE 4.1 Protocol Stack
23. FIGURE 4.2 Part of the Protocol Stack
24. FIGURE 4.3 Infrastructure based architecture
25. FIGURE 4.4 Adhoc Architecture
26. FIGURE 4.5 Network Model for WLAN in OPNET
27. FIGURE 4.6 Access Point Parameters
28. FIGURE 4.7 Subscriber Station Parameters
29. FIGURE 4.8 Average voice jitters without Silence Suppression
30. FIGURE 4.9 Average voice jitters using Silence Suppression
31. FIGURE 4.10 Average packet end to end delay without Silence Suppression
32. FIGURE 4.11 Average packet end to end delay using Silence Suppression
vii
33. FIGURE 4.11 Average MOS without Silence Suppression
34. FIGURE 4.12 Average MOS using Silence Suppression
35. FIGURE 4.13 Network Model for WLAN
36. FIGURE 4.14 Average Jitter
37. FIGURE 4.15 Average Packet end to end delay
38. FIGURE 5.1 Protocol Stack of mobile Internet Device
39. FIGURE 5.1 Network Model for WiMAX-WLAN Static Integrated Network
40. FIGURE 5.2 Protocol Stack of SS_WiMAX_WLAN_AP
41. FIGURE 5.3 Average voice Jitter without Silence Suppression
42. FIGURE 5.4 Average voice Jitter with Silence Suppression
43. FIGURE 5.5 Average voice MOS without Silence Suppression
44. FIGURE 5.6 Average voice MOS with Silence Suppression
45. FIGURE 5.7 Average Packet end-to-end delay without Silence Suppression
46. FIGURE 5.8 Average Packet end-to-end delay with Silence Suppression
47. FIGURE 5.9 Network Model for WiMAX-WLAN Integrated Mobile Network
48. FIGURE 5.10 Average Voice Jitter without silence suppression
49. FIGURE 5.11 Average Packet end-to-end delay without silence suppression
50. FIGURE 6.1. Wireless Network Deployment Wizard
51. FIGURE 6.2. Technology Window
52. FIGURE 6.3. Topology Window, First Tab
53. FIGURE 6.4. Configuration Summary
54. FIGURE 6.5. Subscriber Station Parameters
55. FIGURE 6.6. Deployed WiMAX Scenario
56. FIGURE 6.7. Profile Setting
57. FIGURE 6.8. Application Deployment
58. FIGURE 6.9. SIP Server Attributes
59. FIGURE 6.10. SIP UAC Parameter
60. FIGURE 6.11. Deployed Mobile WiMAX Network
61. FIGURE 6.12. Technology Selection Window
62. FIGURE 6.13. Topology Selection Window
63. FIGURE 6.14. Configuration Summary Window
64. FIGURE 6.15. Deployed WLAN Network
65. FIGURE 6.16. WLAN Profile Configuration
66. FIGURE 6.17. WLAN Application Deployment
viii
67. FIGURE 6.18. Mobile WLAN Network
68. FIGURE 6.19. Static WLAN-WiMAX Integrated Network
69. FIGURE 6.20. Deployed WLAN-WiMAX Integrated Mobile Network
70. FIGURE 7.1. Average voice jitter
71. FIGURE 7.2. Average voice jitter with Silence Suppression
72. FIGURE 7.3. Average packet end to end delay
73. FIGURE 7.4. Average packet end to end delay with Silence Suppression
74. FIGURE 7.5. Average voice MOS in WiMAX and WLAN and Integrated network
75. FIGURE 7.6. Average voice MOS with Silence Suppression
76. FIGURE 7.7. Average Jitter in mobile network
77. FIGURE 7.8. Average Packet End-to-End Delay for three networks
78. FIGURE 7.9. Average Packet End-to-End Delay for two networks
ix
List of Publications
1. “Study of OPNET and performance evaluation of WiMAX network under
various terrain conditions in OPNET”, Proceedings of National Conference
on Microwave and Communication 2010, 19-20 February 2010, Suri, West
Bengal, India.
2. “Comparison of VoIP Performance over WiMAX, WLAN and WiMAXWLAN Integrated Network using OPNET”, Accepted in the International
Conference on Wireless and Mobile Networks 2010, 23-25 July, Chennai,
India.
3. “Reducing Hand-off Latency in WiMAX network using Cross Layer
Information”, accepted in International Conference on Advances in
Computer Engineering 2010, 21-22 June 2010, Bangalore, India.
4. “Performance Study of Mobile WiMAX using Adaptive Modulation and
Coding under varied speed and pathloss”, Communicated.
5. “Comparison of VoIP Performance over WiMAX, WLAN Network using
OPNET”, Communicated.
x
Abbreviations
AAA
Authentication, Authorization and Accounting
ADPCM
Adaptive Differential Pulse Coded Modulation
AES-CCM
Advanced Encryption Standard Counter with
CBC MAC
AMC
Adaptive Modulation and Coding
AP
Access point
ARQ
Automatic Repeat Request
ASN
Access Service Network
AWGN
Adaptive White Gaussian Noise
BE
Best Effort Service
BPSK
Binary Phase Shift Keying
BSS
Base Service Set
BWA
Broadband Wireless access
CBR
Constant Bit Rate
CID
Connection Identifier
CS
Convergence Sub-layer
CS-ACELP
Conjugate Structure Algebraic-Code Excited
Linear Prediction
CSMA/CA
Carrier Sense Multiple Access/ Collision
Avoidance
CSN
Connectivity Service Network
CCA
Clear Channel Assessment
DBPSK
Differential Binary Phase Shift Keying
DCF
Distributed Coordination Function
DCME
Digital Circuit Multiplication Equipment
DHCP
Dynamic Host Control Protocol
DL
Downlink
DLC
Data Link Control Layer
xi
DL-MAP
Downlink Map
DOCSIS
Data over cable service interface specification
DQPSK
Differential Quadrature Phase Shift Keying
DSL
Digital Subscriber Line
DSSS
Direct Sequence Spread Spectrum
EAP
Extensible Authentication Protocol
ertPS
extended real time Polling Service
ESS
Extended Service Set
FDD
Frequency Division Duplexing
FHSS
Frequency Hop Spread Spectrum
FTP
File Transfer Protocol
GFSK
Gaussian Frequency Shift Keying
GRE
Generic Routing Encapsulation
IETF-EAP
Internet Engineering Task Force-Extensible
Authentication Protocol
IEEE
Institute of Electrical and Electronic Engineers
IP
Internet Protocol
IR
Infra Red
ISI
Inter Symbol Interference
ISM
Industrial, Scientific and Medical
ITU-T
Telecommunication Standardization Sector of the
International Telecommunications Union
LAN
Local Area Network
LD-CELP
Low-Delay Code Excited Linear Prediction
LLC
Logical Link Control
LOS
Line Of Sight
MAC
Medium Access Control
MAC CPS
MAC Common Part Sub-layer
MN
Mobile Node
xii
MOS
Mean Opinion Score
MS
Mobile Station
nrtPS
non-real time Polling Service
NSP
Network Service Provider
NWG
Network Working Group
OFDMA
Orthogonal Frequency Division Multiple Access
PC
Point coordinator
PCF
Point Coordination Function
PHY
Physical layer
PLCP
Physical Layer Convergence Protocol
PMD
Physical Medium Dependent
PMKv2
Privacy and Key Management Protocol version 2
PPP
Point to Point Protocol
PSTN
Public Switched Telephone Network
PTM
Point To Multipoint
PTP
Point To Point
QAM
Quadrature Amplitude Modulation
QoS
Quality of Service
QPSK
Quadrature Phase Shift Keying
RLC
Radio Link Control
rtPS
real time Polling Service
RTS/CTS
Request-To-Send/ Clear-To-Send
SDU
Service Data Units
SIP
Session Initiation Protocol
SISO
Single Input Single Output
SONET
Synchronous Optical Network
SS
Subscriber Station
TDD
Time Division Duplexing
xiii
TDM
Time Division Multiplexed
TDMA
Time Division Multiple Access
TEK
Traffic Encryption Key
UGS
Unsolicited Grant Service
UL
Uplink
UL-MAP
Uplink Map
VBR
Variable Bit Rate
VoIP
Voice over IP
Wi-Fi
Wireless Fidelity
WiMAX
Worldwide Interoperability for Microwave
Access
Wireless MAN
Wireless Metropolitan Area Network
Wireless HUMAN
Wireless High-Speed Unlicensed Metropolitan
Area Networks
WLAN
Wireless Local Area Network
WWAN
Wireless Wide Area Network
xiv
Contents
Chapter 1: Introduction……………………………………………..1
1.1 Introduction………….…………………..……………………….2
1.2 Voice over IP (VoIP).……………………………………………4
1.3 Problem Statement….……………………………………………9
1.4 Related Work……….……………………..……………………10
1.5 Thesis outline……….…………………..………………………10
Chapter 2: Effect of Terrain and Pathloss on Wireless Networks.12
2.1 Introduction…………………………………………………..…13
2.2 Background………………………………………………..……13
2.2.1 Free Space Propagation Model……………………………….…...13
2.2.2 Erceg’s Suburban Fixed Model………………………………........14
2.2.3 Outdoor-to-Indoor and pedestrian pathloss Model…………..……16
2.2.4. Vehicular environment…………………………………….…..….16
2.3. Simulation Setup…………………………………………….....16
2.4. Simulation Results and Discussion……………………….……18
2.4.1 Pathloss……………………………………………………..……..18
2.4.2 Average Throughput………………………………………………19
2.4.3 Average Packet end to end delay……………………………...….20
2.5 Conclusion…………………………………………………..….21
Chapter 3: VoIP over WiMAX Network…………………………..22
3.1 Introduction…………………………………………………….. 23
3.2 Background of IEEE 802.16 and WiMAX …………………….23
3.3 Overview of 802.16 MAC Layer………………….……………25
3.4 Overview of 802.16 PHY Layer………………………………..28
3.5 WiMAX Network Architecture…………………...……………29
3.6 Experimental Setup……………………………………………..31
3.6.1 Scenario 1: Simulation Setup and Results………………………....31
3.6.1.1 Average Jitter…………………………………………....32
3.6.1.2 Average Packet End to End Delay……………………...34
3.6.1.3 Average MOS…………………………………………..35
3.6.2 Scenario 2: Simulation Setup and Results…………………………35
3.6.2.1 Average Jitter and packet end to end delay……………..37
3.7 Conclusion…………………………………………...………….38
Chapter 4: Voice over WLAN……………………………...…...….39
4.1 Introduction……………………………………………...……...40
4.2 Background of IEEE 802.11 and WLAN……………….………40
4.3 Protocol Architecture…………………………………………...41
4.4 Overview of IEEE 802.11 MAC Layer…………………………43
4.4.1 IEEE 802.11 MAC: DCF and PCF………………………..………43
4.5 Overview of IEEE 802.11 PHY layer……………………...…...44
4.5.1 Frequency Hop Spread Spectrum (FHSS)…………………………44
4.5.2 Dynamic Sequence Spread Spectrum (DSSS)……………….……44
4.5.3 Infrared……………………………………………………….……45
4.6 WLAN System Architecture………………..……………….….45
4.6.1 Infrastructure mode……………………………………45
4.6.2 Adhoc mode…………………………………………...46
4.7 Experimental Setup……………………………………………..47
4.7.1 Scenario 1: Simulation Setup and Results…………………………47
4.7.1.1 Average Jitter……………………………………………48
4.7.1.2 Average Packet End to End Delay……………………...49
4.7.1.3 Average MOS…………………………………………..50
4.7.2 Scenario 2: Simulation Setup and Results…………………………51
4.7.2.1 Average Jitter……………………………………………52
4.7.2.2 Average Packet End to End Delay……………………...52
4.8 Conclusion………………………………………………………53
Chapter 5: Voice over WiMAX-WLAN Integrated Network……54
5.1 Introduction……………………………………………………..55
5.2 WLAN vs. WiMAX…………………………………………….56
5.3 Types of Integration…………………………………………….58
5.3.1 Loose Coupling……………………………………………………58
5.3.2 Tight Coupling………………………………………………….…58
5.4 Experimental Setup……………………………………………..59
5.4.1 Scenario 1: Simulation Setup and Results…………………………59
5.4.1.1 Average Jitter……………………………………………60
5.4.1.2 Average MOS …………………………………………..61
5.4.1.3 Average Packet End-to-End Delay……………………...62
5.4.2 Scenario 2: Simulation Setup and results………………………….63
5.4.2.1 Average jitter………………………………...………….63
5.4.2.2 Average Packet End-to-End Delay……………………...64
5.6 Conclusion………………………………………………………65
Chapter 6: Development of Scenarios in OPNET ……..…………66
6.1 Why OPNET?..............................................................................67
6.2 Development of WiMAX in OPNET…………………………...67
6.2.1 Static Scenario……………………………………………………..68
6.2.2 Mobile Scenario……………………………………………………76
6.3 Development of WLAN in OPNET…………………………….82
6.3.1 Static Scenario……………………………………………………..82
6.3.2 Mobile Scenario……………………………………………………87
6.4 Development of WLAN-WiMAX Integrated Scenario in
OPNET……………………………………………………………...88
6.4.1 Static Scenario……………………………………………………..88
6.4.2 Mobile Scenario……………………………………………………90
Chapter 7: Comparative Analysis….………………………………92
7.1 Introduction……………………………………………………..93
7.2 Comparative Results……………………………………………93
7.2.1 Stationary network…………………………………………………93
7.2.2 Mobile network……………………………………………………97
7.3 Conclusion………………………………………………………99
Chapter 8: Conclusion and Future Work………………………..100
Bibliography
Chapter 1
Introduction
Chapter Outline: Introduction
 Voice over IP (VoIP)
 Problem Statement
 Related Work
 Thesis Outline
1
Introduction
1.1 Introduction
Wireless networking has become an essential part in the modern
telecommunication system. The demand of high speed data transfer with quality
has led to the evolution of technologies like WiMAX and WLAN and is still
increasing. Hence, new ways to enhance quality and speed of connectivity are
being searched for. With step towards the fourth generation communication
networks, integrated networks are coming into operation. Also voice over IP is
expected to be a low cost communication medium. The voice codecs are big
constraints which affect the quality of the voice in a network. Hence, before real
time deployment of VoIP over a network it is necessary to evaluate the voice
performance over varying networks for various codecs.
WLANs [1] are mostly designed for private wired LANs and have been
enormously successful for data traffic but voice traffic differs fundamentally from
data traffic in its sensitivity to delay and loss [2]. Voice over WLAN is popular,
but maintaining the speech quality is still one of many technical challenges of the
VoIP system. VoIP is spreading rapidly and there is need to support multiple
concurrent VoIP communications but WLAN support handful number of users [3]
[4].
The IEEE 802.11 MAC specifies two different mechanisms, namely the
contention-based Distributed Coordination Function (DCF) [1] and the pollingbased Point Coordination Function (PCF) [1]. The DCF uses a carrier sense
multiple access with collision avoidance (CSMA/CA) scheme for medium access
and the optional four way handshaking request-to-send/clear-to-send [1]
mechanism (RTS/CTS). Incapability of providing differentiation and prioritization
based upon traffic type results in providing satisfactory performance for best-effort
traffic only, but inferior support for QoS requirements posed by real time traffic.
These requirements make the DCF scheme a less feasible option to support QoS
for VoIP traffic. The PCF mode enables the polled stations to transmit data without
contending for the channel. Studies on VoIP over WLAN in PCF mode [5] shows
that the polling overhead is high with increased number of stations in a basic
service set (BSS). This results in excessive delay and poor performance of VoIP
2
under heavy load conditions. Thus, both DCF and PCF have limited support for
real-time applications. Supporting VoIP over WLAN using DCF mode poses
significant challenges, because the performance characteristics of their physical
and MAC layers are much worse than their wired counterparts and hence
considered in our system.
WiMAX (Worldwide Interoperability for Microwave Access) [6][16] on the
other hand is designed to deliver a metro area broadband wireless access (BWA)
service. So, while wireless LAN supports transmission range of up to few hundred
meters, WiMAX system ranges up to 30 miles [6]. Unlike a typical IEEE 802.11
WLAN with 11Mbps bandwidth which supports very limited VoIP connections
[4], an IEEE 802.16 WiMAX with 70Mbps bandwidth [7] can support huge
number of users. These motivations led to study and comparison of the VoIP
quality of service in IEEE 802.11b WLAN and IEEE 802.16 WiMAX network.
IEEE 802.16 support 5 types of service classes, namely UGS (Unsolicited
Grant Service), rtPS (real time Polling Service), nrtPS (non-real time Polling
Service), BE (Best Effort Service), ertPS (extended rtPS service) [8]. UGS
supports fixed-size data packets at a constant bit rate (CBR). It supports real time
applications like VoIP or streaming applications but wastes bandwidth during the
off periods. rtPS supports variable bit rate(VBR) real-time service such as VoIP.
Delay-tolerant data streams such as an FTP is designed to be supported by the
nrtPS. This requires variable-size data grants at a minimum guaranteed rate. The
nrtPS is similar to the rtPS but allows contention based polling. Data streams, such
as Web browsing, that do not require a minimum service-level guarantee is
supported by BE service. BE connections are never polled but receive resources
through contention. ertPS was introduced to support VBR real-time services such
as VoIP and video streaming. It has an advantage over UGS and rtPS for VoIP
applications as it carries lower overhead than UGS and rtPS [9] and hence is
modeled in the system.
Since, WiMAX is expected to create the opportunity to successfully penetrate
the commercial barrier by providing higher bandwidth, establishing wireless
commons becomes an important factor. Also, bandwidth crunch and network
integration are some of the major technical and social challenges regarding the
future of the community-based Wi-Fi networks [10]. According to [10], the
foundation of the WiMAX PTP commons is the process of hot-spot
3
interconnection and integration. Instead of global Internet connectivity, many
current applications and businesses are expected to be better utilized by using the
localized Wi-Fi constellation.
With a step towards the next generation, it is expected that an integrated
network as shown in Figure 1.1, comprising of both the WiMAX and WLAN
network and using mobile nodes with dual stack is expected to provide a better
performance than a similar WiMAX or WLAN network.
FIGURE 1.1. Wi-Fi integration using WiMAX [10].
VoIP has been widely accepted for its cost effectiveness and easy
implementation. A VoIP system consists of three indispensable components,
namely 1) codec, 2) packetizer, and 3) playout buffer. Analog voice signals are
compressed, and encoded into digital voice streams by the codecs. The output
digital voice streams are then packed into constant-bit-rate (CBR) voice packets by
the packetizer. A two way conversation is very sensitive to packet delay jitter but
can tolerate certain degree of packet loss. Hence a playout buffer is used at the
receiver end to smooth the speech by removing the delay jitter.
Quality of noise sensitive VoIP is usually measured in terms of jitter, MOS
and packet end-to-end delay. Perceived voice with zero jitter, high MOS and low
packet end-to-end delay is considered to be the best. With the two competing
wireless networks namely WLAN and WiMAX, this paper analyses the perceived
voice quality as measured using OPNET simulation environment.
1.2 Voice over IP (VoIP)
Voice is analog and is converted to digital format before transmitting over
Internet. This process is called encoding and the converse is called decoding and
both are performed by voice codecs [11]. With bandwidth utilization becoming a
huge concern, voice compression techniques are used [11] to reduce bandwidth
consumption. Voice compression by a codec adds an additional overhead of
4
algorithmic delay. Thus, a codec is expected to provide good voice quality even
after compression, with minimum delay.
Table 1.1 shows the bandwidth requirements of some common codecs. G.711
is the international standard for encoding telephone audio. It has a fixed bit rate of
64kbps. G.723 and G.729 are low bit rate codecs at the expense of high codec
complexity. G.723 is one of the most efficient codecs with the highest compression
ratio and is used in video conferencing applications. G.729 is an industry standard
with high bandwidth utilization for toll-quality voice calls [12]. G.726 uses
ADPCM speech codec standard, and transmits at rates of 16, 24, 32, and 40 kbps.
G.728 officially codes speech at 16 Kbit/s using low-delay code excited linear
prediction [13]. For example, during a call using G.711 as codec, the amount of
data transfer for both uplink and downlink will be 87.2 x 2 = 174.4Kbps = 0.1703
Mbps = 10.21 MB per minute. So, G.711 uses 10.21 Mb/min per VoIP call where
as G.729 uses 0.5MB/min per voice call in the same way.
TABLE 1.1. Bandwidth Requirement of Some Common Codecs [11][14]
Codecs
Algorithm
Bandwidth Ethernet
(Kbps)
Bandwidth
Usage (Kbps)
G.711
PCM(Pulse Code Modulation)
64
G.729
CS-ACELP (Conjugate Structure 8
87.2
31.2
Algebraic-Code Excited Linear
Prediction)
G.723.1
Multi Rate Coder
6.3
21.9
G.723.1
Multi Rate Coder
5.3
20.8
G.726
ADPCM(Adaptive
Differential 32
55.2
Pulse Code Modulation)
G.726
ADPCM(Adaptive
Differential 24
47.2
Pulse Code Modulation)
G.728
LD-CELP
(Low-Delay
Code 16
31.5
Excited Linear Prediction)
5
Figure 1.2. Packet Creation based on voice activity.
Moreover, recently voice codecs are developed to detect talk-spurt [15] and
silence lengths [15] within a conversation. Silence in a communication period
leads to packetization of the background noise and sending it over the network.
This causes bandwidth wastage. Usually, during a conversation we talk 35% of the
time and remain quiet rest of the time [15]. With silence suppression during the
silence period, the codec does not send data as shown in Figure 1.2. This decreases
channel utilisation and thereby saves bandwidth.
Voice communication is noise sensitive. Noise causes the signal to reach
the destination with a lead or lag in the time period. This deviation is called jitter.
Lead causes negative jitter and lag causes positive jitter and both degrade the voice
quality. The time taken by voice to be transmitted from the mouth of the sender to
the ear of the receiver is called packet end-to-end delay. The packet end-to-end
delay should be very less for voice communication. Perceived voice quality is
typically estimated by the subjective mean opinion score (MOS), an arithmetic
average of opinion score. MOS of a particular codec is the average mark given by
a panel of auditors listening to several recorded samples. It ranges from
1(unacceptable) to 5 (excellent). It depends on delay and packet dropped by the
network. The E-model, an analytical model defined in ITU-T recommendation,
provides a framework for an objective on-line quality estimation based on network
performance measurements like delay and loss and application level factors like
low bit rate codecs. The result of the E-model is the calculation of the R-factor
(best case 100 worst case 0) [5].
R = R0 – Is – Id – Ie + A.
(1.1)
Where R0 groups the effects of noise, Is includes the effects of the other
impairments related to the quantisation of the voice signal, Id represents the
impairment caused due to delay, Ie covers the impairments caused by the low bit
6
rate codecs and packet losses. The advantage factor A compensates for the above
impairments under various user conditions. A is 10 for mobile telephony but 0 for
VoIP [5]. R0 is considered to be 94.77 and Is is considered to be 1.43 in OPNET
14.5.A. The relation between MOS and R-factor:
MOS = 1 + 0.035R + 7.10-6R(R – 60) (100 – R)
(1.2)
Packet networks operate on packet switching principles; hence voice in an
IP or WiMAX network would be transmitted to the destination as a collection of
packets where each one might follow different routes, thus arrive at the destination
with different delays. Typically originating and terminating parts would
respectively be the in-house portions of an enterprise network. The part lying in
Internet or some public WAN (i.e. shared medium) is termed the core network.
The packet flow is assumed to be from left to right. For flows in the opposite
direction, the terminology of originating and terminating is interchanged. Unless
stated otherwise, in view of delay calculations, originating and termination parts
are regarded to be identical.
On the originating side, the analog voice signal is digitized into pulse code
modulation (PCM) signals by the voice codecs. Then the PCM samples are
compressed and converted into packet format, thus ready to be sent across the net.
For some network configurations, the edge router may also perform codec and
compression functions. In the subsections below, we attempt to describe various
kinds of delays together with their associated formulation in units of milliseconds
(ms).
o Look Ahead Delay: The codecs conforming to different standards may use
voice blocks of different sizes. The block size used by the G.729 A codec is
10 ms whereas; codecs of G.723.1 have block sizes of 30 ms [16]-[18]. The
present packet is checked while the previous one is being compressed. Note
that such an operation is not possible at the very beginning of transmission.
Thus, a fixed delay of 5 ms, called look-ahead delay, occurs in this section of
packet voice transmission [16]-[17].
o Encoding and Compression Delay: Generally the G.729 A type codec is
used in VoIP having a mean opinion score (MOS) value of 4.2 which almost
provides toll quality [19]. This type of codec has an 8 kbps modulation rate,
hence producing a 10 ms delay for encoding the packet [17]-[24].
7
o Packetization Delay: The time spent for packetization of encoded and
compressed data packets is called packetization delay. To support a good
quality voice call, the packetization delay should be less than 30 ms [17]. The
packetization process commences after storing the packets into a buffer. As
long as the buffering time remains below 10 ms, the compression and
buffering periods will overlap and there will be no additional delays
introduced at the buffer [17]. Otherwise, the buffering time would exceed the
compression time, and the remaining period would add to the total delay
created during the buffering operation.
o Serialization delay (SD): Serialization of the data/voice packets to the core
network is going to be implemented at the ingress point of the originating
network. This creates an extra delay, which may be calculated as [19]
SD= (payload + 48) x 8 / Link Speed
(1.3)
The term “payload” corresponds to the average number of bytes stored in
payload of the packet and 48 indicates the number of bytes added by real
time protocol, user datagram protocol and internet protocol (RTP/UDP/IP)
headers including some optimizations [17], [19].
o Network Delays: An IP packet suffers several other delays while traversing
the whole network. The sources of these delays are explained below:
o Switching Delay: The packets wait for 10 ms at each switch in the
originating network and 1 ms in the core network [17]. The total
switching delay becomes
Switching delay = NRo x 10 + NRc x 1
(1.4)
Where NRo represents the number of routers in the core network
and the NRc represents the number of routers in the core network.
In the simulation, the originating/terminating hop counts are both
taken as 1.
o Propagation Delay (PD): Assuming that a signal propagates with a
velocity of two thirds of the speed of light, each km of propagation
will contribute an extra 5 μs delay. The propagation delay may be
written as
PD = Distance (in km) x 5 x 0.001
(1.5)
o Data Queuing Delay (DQD): The packets are serial, so they are
served sequentially. The waiting time of each packet will be
8
different, depending on the packet size. Also, these queuing delays
will also vary depending on the transmitted IP packet size. Because
of these complications, data queuing delay is not easy to formulate.
In the simulation, IP is assumed to carry only data packets.
Therefore, the voice packet will have the highest priority as there
will be no video packets. In this way the data packets will therefore
jump ahead of the other packets and wait only for the one that is
currently being processed. Even with this assumption an accurate
prediction of queuing delay is still impossible, but may be
simplified to [19].
o Voice/Data Contention Delay (VDCD): Voice data contention
delay is expressed as
VDCD = (MTU + 48) / Link Speed
(1.6)
The term MTU (maximum transferable unit) appears, because in
worst case assumption, the maximum payload size of the data
packet will be equal to MTU.
o Jitter Buffer Delay: Each packet serialized to the network will follow a
different route according to the status of network and routers at that
moment, so each packet will arrive to destination with a different delay
which is proportional to the maximum jitter size.
o Voice Decoding Delay: The decoding time for each packet at the receiver is
given by [19].
Voice Decoding Delay = Encoding Delay
(1.7)
Since a G.729 codec imposes an encoding delay of 10 ms [17]-[23] our
decoding delay will simply be 10 ms for each packet.
The purpose of this modeling is to compare the performance parameters for
the voice codecs considering both with and without silence suppression in WiMAX
802.16d, WLAN 802.11b and their integrated network and thereby show that the
integration provides optimal network capacity and quality of service.
1.3 Problem Statement
With voice over IP coming into existence, the maximum utilisation of the spectrum
has also become a concern. To accommodate maximum number of users with
9
considerably good quality of voice has become a concern of the researchers.
Innovative methods and techniques are coming up each day to suffice the needs.
With the next generation communication networks coming into being our target is
to analyse of the performance of voice over IP over a WLAN-WiMAX integrated
4G network. WLAN is chosen because it is a widely used low cost technology and
WiMAX is expected to outshine the conventional DSL cables for providing
broadband wireless access. Hence, it is expected that an integrated WLANWiMAX
1.4 Related Work
[1], [2], [4], [5], [7] and [13] discusses the performance and capacity of the
WLAN networks while [8], [9], [16] discusses the capacity and performance of
WiMAX network. The idea of the voice codecs has been discussed in [15]. It also
discusses the concept of silence suppression in it. In [10] it is proposed that an
integrated architecture of WiMAX and WLAN is expected to perform better than
the conventional WiMAX or WLAN network. Among them [2], [7] and [13]
discusses the capacity improvement of WLAN network for voice traffic while
quality of the voice traffic has been taken care of in [4] and [5]. Papers [6] [8] [9]
have discussed the performance of the WiMAX network for voice over IP and
multimedia traffic. To the best of our knowledge, the performance of WLANWiMAX integrated network has not been evaluated yet. In this thesis we provide a
comparative discussion of the performance of a WLAN-WiMAX integrated
network with respect to the upcoming application voice over IP.
1.5 Thesis Outline
The thesis is organised as follows:
 Chapter 2 discusses the performance of WiMAX network for various
pathloss models.
 Chapter 3 discusses the performance of a WiMAX network in both static
and mobile conditions for the application voice over IP
 Chapter 4 discusses the performance of a conventional WLAN network in
both static and mobile conditions for the application voice over IP.
10
 Chapter 5 discusses the performance of the WLAN-WiMAX integrated
network in both static and mobile conditions for the voice over IP
application.
 Chapter 6 discusses the simulation setup in OPNET to perform the
simulations in the aforesaid chapters
 Chapter 7 shows the comparative study of the network performances for
VoIP in both static and moving conditions
 Chapter 8 discusses the future work that can be done as an extension to this
thesis and finally concludes the thesis.
11
Chapter 2
Effect of Terrain and Pathloss on
Wireless Networks
Chapter Outline: Introduction
 Background
 Free Space Propagation Model
 Erceg’s Suburban Fixed Model
 Outdoor-to-Indoor and pedestrian pathloss Model
 Vehicular environment
 Simulation Setup
 Simulation Results and Discussion
 Pathloss
 Average Throughput
 Average Packet end to end delay
 Conclusion
12
Effect of Terrain and Pathloss on
Wireless Networks
2.1. Introduction
With various wireless technologies coming up, performance of the wireless
networks has become a concern. Mobile WiMAX is expected to be the wireless
technology of the next generation as it supports vehicular mobility with broad
coverage area. Among various other parameters which affect network
performance, the terrain feature on which the network is deployed and the path loss
model affects the network performance substantially. Air being the communicating
medium in wireless networks, any hindrance on the communication path like
buildings, trees, etc., affects the wireless signal propagation. Hence, prior network
deployment, analysis of the terrain is essential. Voice over IP as discussed in
Chapter 1 is expected to be the communication medium of the next generation and
File Transfer Protocol is one of the most popular data transfer protocol. Hence, in
this chapter the behaviour of various propagation models in WiMAX network with
respect to path loss, throughput and delay for various terrain models with voice and
FTP applications is studied.
2.2. Background
The common propagation models namely Free Space path loss model,
Suburban Fixed (Erceg), Outdoor to Indoor and Pedestrian Environment and
Vehicular Environment are discussed in this paper. The Suburban Fixed path loss
model is further subdivided in three categories based on the building and tree
density namely terrain type A, B and C. The models are briefly discussed in the
next sections.
2.2.1. Free Space Propagation model
The free space propagation model is mathematically given by:
13

(2.1)
Where, Prx is received power in watts and is a function of distance between
transmitter and receiver, Ptx is the transmitted power in watts, Grx and Gtx are the
gain of the receiving and transmitting antennas respectively, L is the system-loss
factor and is not related to propagation. It is usually greater than 1 and λ is the
wavelength in meters [25].
2.2.2. Erceg’s Suburban Fixed Model
The Erceg model is based on extensive experimental data collected at
1.9GHz in 95 macro cells of suburban areas across the United States [26]. This
model is a slope intercept model given by [26]:
(2.2)
Where, PL is the instantaneous attenuation, H is the intercept and is given by free
space path loss at the desired frequency over a distance of d0 = 100 meters:

Where, λ is the wavelength. The parameter
(2.3)
is a Gaussian random variable over
the population of macro cells within each terrain category. It can be written as [26]
[27].
(2.4)
where hb is the height of the base station antenna in meters, σγ is the standard
deviation of , x is a zero-mean Gaussian variable of unity standard deviation N[0,
1], and e, g, k and σγ are all data-derived constants for each terrain category. The
shadow fading components varies randomly from one terminal location to another
within any given macro-cell. It is a zero-mean Gaussian variable and can be
expressed as [26] [27].

(2.5)
14


where y and z are the zero-mean Gaussian variables of unit standard deviation N[0,
1], σ is the standard deviation of s, µσ is the mean of σ, and σσ is the standard
deviation of σ. µσ and σσ are both data-derived constants for each terrain category.
The numerical values of the above parameters are given in Table 2.1.
The correlation factors of the model for the operating frequency and for the
MS antenna height are given in [27] as:

for Terrain type A and 

for Terrain type C
Where, f is the frequency in MHz and hr is the height is the MS antenna above
ground in meters.
The Terrain Type A is a hilly terrain with moderate to heavy tree density,
representing rural environments and has highest path loss. Terrain Type B is
characterized by either a mostly flat terrain with moderate to heavy tree density or
a hilly terrain with light tree density. Terrain Type C is a flat terrain with light tree
density and is associated with the lowest path loss for rural environments [27].
TABLE 2.1. NUMERICAL VALUES CONSIDERED FOR THE PARAMETERS [26]
Parameters
 
e
g m 1
k m 



Terrain
Terrain
Terrain
Type A
Type B
Type C
4.6
4.0
3.6
0.0075
0.0065
0.005
12.6
17.1
20.0
0.57
0.75
0.59
10.6
9.6
8.2
2.3
3.0
1.6
15
2.2.3. Outdoor-to-Indoor and Pedestrian Pathloss Model
This environment is characterized by small cells and low transmission
power. Base stations with low antenna heights are located outdoors; pedestrian
users are located on streets and inside buildings and residences.
(2.9)
The above equation describes the path loss in dB where R is the distance
between the base station and the mobile station in kilometres and ‘f’ is the carrier
frequency of 2000 MHz for IMT-2000 band application [28].
2.2.4. Vehicular environment
This environment is characterized by larger cells and higher transmits power. The
path loss in vehicular environment in dB is given by:
PL = 40(1 – 4 X 10-3Δhb) log10 R – 18 log10 Δhb
– 21 log10 f + 80dB
(2.10)
where, R is the distance between the base station and the mobile station, and f is
the carrier frequency of 2000 MHz and Δhb is the base station antenna height in
meters measured from the average roof top level.
2.3. Simulation Setup
The Wireless Deployment Wizard of OPNET is used to deploy a 7 celled
WiMAX network, with multiple subscriber stations in the range of a base station as
shown in Figure 2.1. The base stations are connected to the core network by an IP
backbone. There is a server backbone containing the voice server which is
configured as the SIP server. The IP backbone is connected to the server backbone
via an ASN gateway. This node, configured as the ASN gateway supports the
mobility in the WiMAX network. The IP backbone, server backbone and the ASN
gateway together represent the service provider company network. The green
bidirectional dotted lines represent the generic routing encapsulation (GRE)
tunnels. The cell radius is set to 30 kilometres. There are 10 nodes under the base
station 2. 5 of these nodes are communicating with 5 mobile nodes under base
station 7 and other 5 are communicating with 5 mobile nodes under base station 3.
Voice call of PSTN quality is configured between these mobile nodes. The nodes
participating in the same session are connected by the blue bidirectional dotted
lines. The mobile nodes under base station 2 are configured to move at a speed of
50 km/hr in the path as shown by the white lines in Figure 2.1. The remaining key
16
network configuration parameters in OPNET are summarized as in Table 2.2 and
the attributes of the network components are shown in Table 2.3.
FIGURE 2.1. Network Model for WiMAX.
TABLE 2.2. NETWORK CONFIGURATION DETAILS
Network
7 celled WiMAX network
Cell Radius
30km
No. of Base Stations
7
No. of Subscriber Stations per BS
10
No. of Mobile nodes in the network 10
Speed of the mobile nodes
50 km/hr
Simulation time
600 sec
Base Station Model
wimax_bs_ethernet4_slip4_router
Subscriber Station Model
wimax_ss_wkstn
ASN Gateway Model
ethernet4_slip8_gtwy
IP Backbone Model
ip32_cloud
Voice Server Model
ppp_server
Link Model (BS-Backbone)
PPP_DS3
Link Model (ASN - Backbone)
PPP_SONET_OC12
17
TABLE 2.3. ATTRIBUTES OF THE NETWORK COMPONENTS
Attributes
Value
Physical Layer Model
OFDMA 20Mhz
MAC Protocol
IEEE 802.16e
Multipath Channel Model
ITU Vehicular A
No. of Transmitter per BS
SISO
Traffic Type of Service
Interactive Voice and Data
Scheduling Type
ertPS, nrtPS
Application
Voice (PSTN quality), FTP
Voice Codec
G 711
FTP Load
High
2.4. Simulation Results and Discussion
The performance metric used to analyse the network performance is path
loss, average throughput of the WiMAX network and packet end to end delay.
Using OPNET, the simulation is conducted for the common path loss models and
the three types of terrains: Terrain Types A, B, C of erceg path loss model. We are
interested to investigate the combined effect of terrain (buildings, trees, etc.) and
vehicular mobility on the Mobile WiMAX performance in OPNET. To investigate
the combined effect of terrain and vehicular mobility, the terrains are simulated by
choosing the terrain type in OPNET which is thereby selected by the simulator,
based on the actual location and surrounding terrain of the transmitter-receiver
pair.
2.4.1. Pathloss
Path loss models are broadly classified into four groups namely free space,
Suburban fixed, Outdoor to indoor and pedestrian, vehicular environment. The
path loss due to these models in a WiMAX network with mobile nodes moving at a
speed of 50 km/hr is modeled in OPNET modeler and the result is shown in Figure
2.2.
18
FIGURE 2.2. Path loss due to various pathloss models in decibel.
From the graph it is observed that the path loss for outdoor to indoor and
pedestrian is the highest and the same for free space is the lowest. This is due to
the fact that the path loss value varies with the amount of reflection in the
communicating path. As a mobile node moves from outdoor to indoor the number
of reflections in the communicating path increases rapidly thereby causing huge
path loss. Also, we see that the path loss for free space is the least. This is
concluded in [27] that the path loss increases with the number of obstructions in
the communication path. Free Space implies a terrain with no or very less
obstruction. Hence, number of obstruction decreases and pathloss thereby is
minimum.
2.4.2. Average Throughput
This metric measure the amount of voice traffic and FTP traffic received in
bits per second on an average for each connection. From Figure 2.3, it is observed
that the throughput of the network with free space path loss model is the highest
and the same for outdoor to indoor and pedestrian is the lowest for both voice and
FTP traffic. This is due to the fact that as the density of obstacles increases, the
Line of Sight (LOS) gets affected. Thus the number of times the signal gets
obstructed and reflected is more. This results in increasing attenuation and
diffraction due to the building structures, trees or mountains. As the Line of Sight
(LOS) between the transmitting and receiving nodes decreases, it causes delay.
This results in packet loss thereby causing fall in the average throughput.
19
FIGURE 2.3. WiMAX Network Throughput in Bits per second
2.4.3. Average Packet end to end delay
This metric measures the time taken by each voice packet to travel from the
mouth of the transmitter to the ear of the receiver and each FTP packet from the
application layer of the sender to the application layer of the receiver, on average
for each connection. From Figure 2.4, it is observed that the packet end to end
delay of the network with free space as the pathloss model is the lowest and the
same for outdoor to indoor and pedestrian is the highest irrespective of the
application. This is due to the fact that as the density of the building structures,
trees or mountains increases, the number of times the signal gets obstructed and
reflected is more thereby increasing attenuation and diffraction. Hence, the LOS
between the transmitting and receiving nodes decreases. This causes the mobile
node to get disconnected from the base stations and thereby perform re-registration
[1] to get re-connected to a base station.
The registration procedure is time consuming and hence adds up to the
packet end to end delay. With flat terrain model and very less or nil obstructions,
the mobile node remains connected to some or the other base stations and thus the
network scanning and registration process is avoided thereby causing less delay as
shown in Figure. 2.4.
20
FIGURE 2.4. Packet end to end delay in seconds
2.5. Conclusion
The target of this work was to check the variation of the network
performance due to varying pathloss and terrain models. This work shows the
variation in the WiMAX network performance with varying path loss models
deployed over suburban areas of terrain types varying from hilly terrain with
moderate to heavy tree density to flat terrain with light tree density for Voice over
IP and File Transfer Protocol application with mobile nodes moving at a speed of
50 km/hr. The aim was to evaluate the effect of various path loss models over
varying suburban terrains, on the basis of average throughput and packet end to
end delay and path loss of Mobile WiMAX. The reduction in Line-Of-Sight due to
the terrain model of any area influences the network throughput directly by
increasing the attenuation and diffraction losses, and indirectly affects the packet
end to end delay by causing nodes to initiate network re-registration more
frequently under fluctuating cell coverage. This is already mentioned in [26] [27]
[29]. We conclude that our results tally with the already taken measurements of the
various path loss models in [26] [27] [29] and thus, free space pathloss model is
chosen for the future works as it is a basic pathloss models with all other
parameters related to terrain and building density set as constant.
21
Chapter 3
VoIP over WiMAX Network
Chapter Outline: Introduction
 Background of IEEE 802.16 and WiMAX
 Overview of 802.16 MAC Layer
 Overview of 802.16 PHY Layer
 WiMAX Network Architecture
 Experimental Setup
 Scenario 1: Simulation Setup and Results
 Average Jitter
 Average Packet End to End Delay
 Average MOS
 Scenario 2: Simulation Setup and Results
 Average Jitter
 Average Packet End to End Delay
 Conclusion
22
VoIP over WiMAX Network
3.1. Introduction
In recent years, internet access has moved to a new dimension. It is now not
restricted to Web browsing and emailing. Multimedia services including Voiceover-IP (VOIP) and media streaming have become the expectation of the next
generation.
To provide users with such connectivity, Broadband Wireless
Access (BWA) comes into the picture. It promises users to be provided with
megabit internet access seamlessly. One of the many technologies under BWA is
WiMAX (Worldwide Interoperability for Microwave Access). Based on IEEE
802.16 it has been designed to provide metro area broadband wireless access. With
70 Mbps speed and over 50 miles of coverage area [30], WiMAX supports
mobility up to 70-80 miles/hr and is expected to be the replacement of cable and
Digital Subscriber Line (DSL) [31]. Hence, with an overview of the IEEE 802.16
BWA network, a discussion of the quality of voice achieved by using a WiMAX
network supporting both in static and moving nodes is done in this chapter.
3.2. Background of IEEE 802.16 and WiMAX
The IEEE 802.16 group was formed in 1998 to develop an air interface
standard for wireless broadband. The group’s initial focus was the development of
a LOS based point-to-multipoint wireless broadband system for operation in the
10-66 GHz millimeter waveband. The resulting standard – the original IEEE
802.16 standard completed in December 2001 – was based on a single carrier
physical (PHY) layer with a burst time division multiplexed (TDM) MAC layer.
Many of the concepts related to the MAC layer were adapted for wireless from the
popular cable modem DOCSIS (data over cable service interface specification)
standard.
The IEEE 802.16 group subsequently produced 802.16a, an amendment to
the standard, to include NLOS applications in the 2 GHz – 11 GHz band using
orthogonal frequency division multiple access (OFDMA), were also included.
Further revisions resulted in a new standard in 2004, called IEEE 802.16d – 2004
which replaced all prior versions and formed the basis for the first WiMAX
23
solution. These early WiMAX solutions based on IEEE 802.16d – 2004 targeted
fixed applications, and are referred to as Fixed WiMAX [32]. In December 2005,
the IEEE Group completed and approved IEEE 802.16e – 2005, an amendment to
the IEEE 802.16d – 2004 standard that added mobility support. The IEEE 802.16e
– 2005 forms the basis for the WiMAX solution for nomadic and mobile
applications and is often referred to as mobile WiMAX [33]. The basic
characteristics of various IEEE 802.16 standards are summarized in Table 3.1.
TABLE 3.1. Basic Characteristics of various IEEE 802.16 standards.
802.16
Completed
Status
December
2001
802.16d – 2004
802.16e – 2005
Completed
Completed June
December
2004
2005
10GHz-66GHz for
fixed;
Frequency
10GHz-66GHz
Band
2GHz-11GHz
2GHz-11GHz for
mobile
applications
Application
Fixed LOS
Point-to-
MAC
Architecture
Multipoint,
mesh
Fixed and Mobile
Fixed NLOS
NLOS
Point-to-Multipoint,
Point-to-
mesh
Multipoint, mesh
Single carrier, 256
Transmission
Single Carrier
Scheme
only
Single Carrier, 256
OFDM
or
OFDM
2048
OFDM
OR
Scalable
OFDM
with
128,
512,
1024,
or
2048
subcarriers
Modulation
Gross
Rate
QPSK,16QAM, QPSK, 16QAM, 64 QPSK,
64QAM
Data 32Mbps134.4Mbps
16QAM,
QAM
64 QAM
1Mbps – 75Mbps
1Mbps – 75Mbps
24
Multiplexing
Duplexing
Burst
Burst TDM/ TDMA/ Burst
TDM/TDMA
OFDMA
TDMA/ OFDMA
TDD and FDD
TDD and FDD
TDD and FDD
1.75 MHz, 3.5 MHz,7
Channel
Bandwidths
20MHz,
MHz, 14 MHz, 1.25
25MHz,
MHz, 5 MHz, 10
28MHz
MHz, 15 MHz, 8.75
MHz
WirelessMAN
Air-interface
WirelessMAN
OFDM
designation
-SC
WirelessMAN–
–
3.5
MHz,7 MHz, 14
MHz, 1.25 MHz,
5 MHz, 10 MHz,
15
MHz,
8.75
SCa
WirelessMAN
WirelessMAN
–
–
OFDMA
WirelessHUMAN
256 – OFDM as Fixed
WiMAX
–
OFDM
WirelessHUMAN
None
MHz,
WirelessMAN
OFDMS
implementation
1.75
MHz
WirelessMAN – SCa
WiMAX
TDM/
Scalable OFDMA
as
Mobile
WiMAX
3.3. Overview of 802.16 MAC Layer
IEEE 802.16 Medium Access Control (MAC) generally follows point to
multi-point (PMP) network topology with support for mesh topology.
From the reference model as shown in Figure 3.1 there are three sub layers
in the MAC:

Service
Specific
Convergence
Sub-Layer
(CS):
providing
any
transformation or mapping of external network data through CS SAP (CS

Service Access Point)
MAC Common Part Sub-layer (MAC CPS): classifying external network
service data units (SDUs) and associating these SDUs to proper MAC
service flow and connection identifier (CID). Multiple CS specifications

are provided to interface with various protocols.
Privacy (or Security) Sub-layer: supporting authentication, secure key
25
exchange, and encryption.
FIGURE 3.1. Reference Model of IEEE 802.16
Unlike the random access techniques of typical MACs of IEEE 802, IEEE
802.16 MAC is connection oriented and similar to time division multiple access
(TDMA). As a subscriber station (SS) enters the network, it creates one or more
connections with the base station (BS). It also performs link adaptation and
automatic repeat request (ARQ) functions to maintain the target bit error rate.
Further to support multimedia services, the IEEE 802.16 MAC may have to use
radio resources and provide quality of service (QoS) differentiation in services
which are not considered typical MAC functions. The five different types of
services to support QoS differentiation for different applications provided by IEEE
802.16 are as follows [35]:
o Unsolicited Grant Services (UGS): UGS is designed to support fixed-size
data packets at a constant bit rate (CBR). It supports real time applications
like T1/E1 emulations but wastes bandwidth during the off periods or
silence periods. Hence, it cannot be used for Voice over IP without silence
suppression.
o Real-Time Polling Services (rtPS): It is designed to support variable sized
data packets, periodically like MPEG video and Voice over IP with silence
suppression.
o Non-Real-Time Polling Services (nrtPS): nrtPS is designed to support non
real time services that require data bursts of variable size on a regular basis
like FTP. It is similar to rtPS except for the contention based polling
scheme.
26
o Best Effort (BE): Data streams that do not require a minimum service-level
guarantee is supported by BE service. It counts typical data traffic like
Internet web browsing and FTP file transfer.
o Extended Real-Time Polling Service (ertPS): It is designed to support
variable rate real-time services such as Voice over IP and video streaming.
It has an advantageous over UGS and rtPS for VoIP applications because it
carries lower overhead than UGS and rtPS [9] and hence is considered in
our system.
To support OFDMA PHY, the MAC layer is responsible for assigning frames
to the proper zones and exchanging this structure information with the SSs in the
DL and UL maps. IEEE 802.16 MAC is connection oriented. As BS controls the
access to the medium, bandwidth is granted to the SSs on demand. At the
beginning of each frame, the BS schedules the uplink and downlink grants to meet
the negotiated QoS requirements. Each SS learns the boundaries of its allocation
under current uplink and sub-frame via the UL-MAP message. The DL-MAP
delivers the timetable of downlink grants in the downlink sub frame.
There are different bandwidth request schemes in WiMAX. For UGS, a
fixed amount of bandwidth is requested periodically at the set-up phase of uplink
and no bandwidth is requested explicitly after that. The unicast poll allocates
necessary bandwidth for a polled uplink connection. The broadcast polls are issued
by the BS to all uplink connections, while a truncated exponential back-off
algorithm is employed to resolve possible collisions in polling. Based on the
bandwidth requested and granted, the BS uplink scheduler estimates the residential
backlog at each uplink connection, and allocates future grants. An SS scheduler
must be implemented with each SS MAC, in order to re-distribute the granted
capacity to all its connections. However, note that IEEE 802.16 does not specify
scheduling algorithms that are left to manufacturers.
Similar to the concept of cellular layer - 2/3, IEEE 802.16 MAC has the
radio link control (RLC) to control PHY transition from one burst profile to
another, in addition to traditional power control and ranging.
Another important sub-layer in the IEEE 802.16 MAC is the security sublayer, and an improved version has been developed for the IEEE 802.16e. The
Privacy and Key Management Protocol version 2 (PMKv2) is the basis of Mobile
WiMAX security. Device and user authentication adopts IETF EAP protocol. The
27
traffic encryption follows the IEEE 802.11i using AES-CCM to protect traffic data.
The keys used to derive the cipher text are generated from the EAP authentication.
To avoid further attacks and hostile analysis, a periodic key (TEK) refreshing
mechanism enables improved protection. A three-way handshake scheme in
Mobile WiMAX optimizes the re-authentication mechanism for fast handover by
preventing man-in-the-middle-attacks.
To deal with mobility in Mobile WiMAX, IEEE 802.16e the MAC
specifies MAC layer handover procedure, while the exact handover decision
algorithm is not specifically defined.
Further details regarding the IEEE 802.16 MAC are discussed in [30] [32] [33]
[35] [9] [36] [37].
3.4. Overview of 802.16 PHY Layer
The Figure 3.2 shows the schematic representation of WiMAX
PHY layer. The diagram has basically three blocks which are
o Transmitter,
o Receiver and
o Channel.
FIGURE 3.2. Schematic Model of WiMAX PHY Layer
The transmitter consists of channel encoding, digital modulations, serial to
parallel conversion block followed by an OFDM modulation block and a parallel
to serial conversion block has been shown that has the specific purpose of
serializing the data bits.
The channel encoding block has “block encoder”, “convolutional encoder”
and “puncture vector block” associated with it. The redundancy added to the data
sequence is the sole purpose of the encoder block. In WiMAX scenario, we can
28
only opt for BPSK, QPSK, 16-QAM and 64-QAM modulation. WiMAX PHY uses
Adaptive Modulation and Coding (AMC) which takes into account the channel
SNR to dynamically select the proper modulation technique appropriate for that
channel condition to deliver the maximum throughput. Serial to parallel conversion
of the incoming bit becomes the critical set of operation that determines the
parallel transmission of data bits. OFDM modulation being the key multiplexing
technique in WiMAX helps the transmission of data bits at a very high rate at a
negligible amount of Inter Symbol Interference (ISI) with minimal amount of
packet loss and bit error.
The data transmitted through wireless channel reaches the receiver. The
channel which might be AWGN, Rayleigh etc. determines the effective channel
impairment introduced in the receiver. The receiver section does exactly the
opposite to the transmitter.
3.5. WiMAX Network Architecture
The IEEE 802.16e – 2005 standard provides the air interface for WiMAX
but does not define the full end-to-end WiMAX network. The WiMAX Forum's
Network Working Group (NWG) is responsible for developing the end-to-end
network requirements, architecture, and protocols for WiMAX, using IEEE
802.16e – 2005 as the air interface. The WiMAX NWG has developed a network
reference model to serve as an architecture framework for WiMAX deployments
and to ensure interoperability among various WiMAX equipment and operators.
The network reference model envisions unified network architecture for
supporting fixed, nomadic, and mobile deployments and is based on an IP service
model. Below is simplified illustration of IP-based WiMAX network architecture.
The overall network may be logically divided into three parts:
o Mobile Stations (MS) used by the end user to access the network.
o The Access Service Network (ASN), which comprises one or more base
stations and one or more ASN gateways that form the radio access network
at the edge. The ASN gateway typically acts as a layer 2 traffic aggregation
points within an ASN. Additional functions that may be part of the ASN
gateway include intra-ASN location management and paging, radio
resource management and admission control, caching of subscriber profiles
and encryption keys, AAA client functionality, establishment and
29
management of mobility tunnel with base stations, QoS and policy
enforcement, and foreign agent functionality for mobile IP, and routing to
the selected CSN.
o Connectivity Service Network (CSN), which provides connectivity to the
Internet, ASP, other public networks, and corporate networks. The CSN is
owned by the NSP and includes AAA servers that support authentication
for the devices, users, and specific services. The CSN also provides per
user policy management of QoS and security. The CSN is also responsible
for IP address management, support for roaming between different NSPs,
location management between ASNs, and mobility and roaming between
ASNs.
The network reference model developed by the WiMAX Forum NWG defines
a number of functional entities and interfaces between those entities. Figure 3.3
below shows some of the more important functional entities.
o Base station (BS): The BS is responsible for providing the air interface to
the MS. Additional functions that may be part of the BS are micro mobility
management
functions,
such
as
handoff
triggering
and
tunnel
establishment, radio resource management, QoS policy enforcement, traffic
classification, DHCP (Dynamic Host Control Protocol) proxy, key
management, session management, and multicast group management.
FIGURE 3.3. WiMAX Network Architecture [39]
30
3.6. Experimental Setup
To analyze VoIP in a network, it is necessary to study real life scenarios.
Hence, OPNET 14.5.A is chosen as the simulation tool as it will reflect the actual
deployment of the WiMAX network.
3.6.1. Scenario 1 – Static Nodes
Figure 3.4 shows the simulation setup used for WiMAX network. Using the
Wireless Deployment Wizard of OPNET a 7 celled WiMAX network, with
multiple subscriber stations in the range of a base station is deployed. The base
station is connected to the core network by a server backbone via an IP backbone.
The server backbone is further connected to the voice server which is configured as
the SIP server. The base station, IP Backbone, Server Backbone and the Voice
Server together represent the service provider company network. The cell radius is
set to be 30 km. The Base Station transmission power is set to 10W and the same
for subscriber station is set to 0.5W based on [35] as shown in Figures 3.5 and 3.6
respectively. The number of subscribers in cells 2 and 3 are 10 and Voice over IP
calls are setup between them in mesh using SIP.
FIGURE 3.4. Network Model for WiMAX in OPNET
31
FIGURE 3.5. Base Station Parameters
FIGURE 3.6. Subscriber Station Parameters
With the simulation setup as mentioned before, the voice codecs being used
for the Voice over IP calls are varied and the corresponding variation in voice
jitter, MOS (Mean Optimal Score) and Packet End-to-end delay are noted.
3.6.1.1. Average Jitter
Figure 3.7 shows the variation of jitter for the WiMAX network
without using silence suppression for various codecs. Perceived voice quality is
best if the jitter is zero. As shown in the figure, average voice jitter is almost 0 for
the voice codecs G 723.1 with both 5.3Kbps and 6.3Kbps and G 726 with 32Kbps
implying very good quality of voice whereas all other codecs shows some
deviation. A positive jitter of 0.000000925926 seconds is shown by G 711 while
all others show negative jitter of about 0.000001841621 seconds. Since, the bit rate
of G 723.1 is 6.3 or 5.3 Kbps; it results in generation of small packets. As stated in
[15] modem and fax signals cannot be carried by G 723.1 and hence it can be used
only for narrow band communications. Like G 711, G 726 has its roots in the
PSTN network. Hence, it is expected to provide users with good quality of voice. It
32
is primarily used for international trunks to save bandwidth. Unlike G 711, G 726
uses 32Kbps to provide nearly the same quality of voice. This is because 32 Kbps
is its de facto standard [34].
To increase the number of users supported count, silence suppression
technique is important. As mentioned in Chapter 1, human speech during one way
conversation has 35% talk spurt length and 65% silence length. Silence
suppression prevents the packetization of the silence length and thereby save
bandwidth. But usage of silence suppression technique increases the positive jitter
considerably compared to the other voice codecs. This is shown in Figure 3.8.
Hence, G 726 cannot be used in cases where silence suppression technique is used
irrespective of its performance in cases where silence suppression has not been
used.
Mean Jitter without silence suppression (sec)
0.0000015
Time(sec)
0.000001
0.0000005
0
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
-0.0000005
-0.000001
-0.0000015
-0.000002
Voice Codecs
FIGURE 3.7. Average voice jitter without Silence Suppression
Mean Jitter with Silence Suppression (sec)
0.03
Time(sec)
0.025
0.02
0.015
0.01
0.005
0
-0.005
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
Voice Codecs
FIGURE 3.8. Average voice jitter using Silence Suppression
33
3.6.1.2. Average Packet End to End Delay
As stated in Chapter 1, packet networks operate on packet switching
principles; hence voice in an IP or WiMAX network would be transmitted to the
destination as a collection of packets where each one might follow different routes,
thus arrive at the destination with different delays. Factors affecting packet end to
end delay include Look Ahead Delay, Packetization Delay, Serialization Delay,
Network Delay, etc. (Details covered in Chapter 1). Figure 3.9 and 3.10 shows, the
Packet End-to-End delay for the voice codec G 723.1 is the highest irrespective of
silence suppression. This is because; G 723.1 uses coding rate of 5.3 Kbps or 6.3
Kbps which results in the formation of packets of smaller size and larger count.
Now as the number of packets increases in the network, the congestion in the
network increases. Congestion directly affects the network packet delay and thus
results in increased packet end to end delay.
Mean Packet-End-to-End Delay without Silence Suppression
0.14
Time (sec)
0.12
0.1
0.08
0.06
0.04
0.02
0
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
Voice Codecs
FIGURE 3.9. Average packet end to end delay without Silence Suppression
Mean Packet End-to-End Delay with Silence Suppression
0.14
Time (sec)
0.12
0.1
0.08
0.06
0.04
0.02
0
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
Voice Codecs
FIGURE 3.10. Average packet end to end delay using Silence Suppression
Also the Look Ahead delay of G 723.1 voice codec is 7.5 msec [35] while
the same for G 729A is 5 msec and other considered voice codecs is 0 msec. On
the other hand the serialization delay of G 726 32Kbps is quite high. As mentioned
in Chapter 1 the serialization delay is directly proportional to the payload size and
is inversely proportional to the link speed. Since the test network has same link
34
configuration, the link speed is same for all codecs. Hence, depending on the
payload size, the variation in the serialization delay affects the total packet end to
end delay significantly.
3.6.1.3. Average MOS
MOS
Mean MOS without Silence Suppression
4
3.5
3
2.5
2
1.5
1
0.5
0
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
Voice Codecs
FIGURE 3.11. Average MOS without Silence Suppression
MOS
Average MOS with Silence Suppression
4
3.5
3
2.5
2
1.5
1
0.5
0
G729
G711
G723_1_5_3 G723_1_6_3
G726_16
G726_24
G726_32
G726_40
G728_12_8
G728_16
Voice Codecs
FIGURE 3.12. Average MOS with Silence Suppression
The Mean Optimal Score (MOS) as shown in Figures 3.11 and 3.12 is
independent of Silence Suppression. As mentioned in Chapter 1, MOS depends on
number of packets dropped. G 723.1 is a low bit rate codec which generates
packets of size 5.3 Kbps or 6.3 Kbps. This results in network congestion and
thereby packet drop. Hence, the MOS value for the voice codec G 723.1 is quite
low. Voice having MOS of 3 can be considered to be of considerable quality.
Hence, all other codecs considerable with respect to their MOS value.
3.6.2. Scenario 2 – Mobile Nodes
Figure 3.13 shows the simulation setup used for WiMAX network.
Using the Wireless Deployment Wizard of OPNET a 7 celled WiMAX network,
with multiple subscriber stations in the range of a base station is deployed. The
35
base station is connected to the core network by a server backbone via an IP
backbone and an ASN Gateway which controls the mobility of the mobile nodes.
The server backbone is further connected to the voice server which is configured as
the SIP server. The base station, IP Backbone, Server Backbone, ASN Gateway
and the Voice Server together represent the service provider company network.
Generic Routing Encapsulation (GRE) tunnel is setup between the ASN
gateway and the base stations to control the mobility of the mobile nodes (for
further details refer to Appendix A). The cell radius is set to be 30 km. The mobile
nodes are configured to move at a rate of 50 km/hr, 100 km/hr and 150 km/hr. The
Base Station transmission power is set to 10W and the same for subscriber station
is set to 0.5W based on [35]. The number of subscribers in cells 2 is 10 and the
same in cells 3 and 7 are 5 each. Voice over IP calls are setup between them using
SIP and are shown by the blue lines in the Figure 3.13.
With the simulation setup as mentioned before, the voice codecs being used
for the Voice over IP calls are varied and the corresponding variation in voice jitter
and Packet End-to-end delay are noted.
Due to the limitation of OPNET, Mean Optimal Score (MOS) could not be
estimated. Also only some of the voice codecs could be modeled. G 723.1 with 5.3
kbps, G 726 with 32 Kbps and G 728 with 16 Kbps could only be modeled.
FIGURE 3.13. Network Model for WiMAX
36
3.6.2.1. Average Jitter
Mean Jitter(sec)
0.02
0.018
0.016
Jitter(sec)
0.014
0.012
50km/hr
0.01
100km/hr
0.008
150km/hr
0.006
0.004
0.002
0
G711
G723.1
G726
G728
G729
Voice Code c
FIGURE 3.14. Mean Jitter without Silence Suppression
Average Packet end to end delay
0.09
0.08
Packet Delay(sec)
0.07
0.06
50km/hr
0.05
100km/hr
0.04
150km/hr
0.03
0.02
0.01
0
G711
G723.1
G726
G728
G729
Voice Codecs
FIGURE 3.15 Mean Packet end to end delay without Silence Suppression
G 711 codec has packet rate of 64 Kbps which is quite large compared to
that of G 723.1 which is about 5.3 Kbps. Thus, G711 has fewer number of packets
compared to G 723.1 for fixed amount of voice. For larger packets the overhead
due to header is small [38]. As a result, for scenario 1, the delay for G 711 is lesser
compared to G 723.1. Moreover, in scenario 1, the nodes are fixed hence the
packets follow almost the same route and arrive in order more or less but wireless
networks, have a inherent property of discarding any packets containing one or
more erroneous bits. So the probability of discarding G723.1 packets are less and
as a result the jitter is also minimum for this codec.
However in the mobile scenario, scenario 2, the probability of packet loss is
more and any packet loss in the physical layer is interpreted as delay by the upper
layers due to the hiding property of the MAC or link layer protocols [38]. Since,
37
the packet size of G 711 is larger; it has higher probability of suffering from packet
loss than G 723.1 thereby having higher delay than G 723.1. This is shown in
Figure 3.15. As the subscriber stations become mobile, their supporting base
stations changes frequently due to handover and the mobile nodes get distributed
between different base stations depending upon their trajectory. Hence, the packets
may follow different path and arrive out-of-order. Since, the number of packets for
G 723.1 is much more than G 711, this problem is more severe than G 711. As a
result it suffers from maximum jitter. This is shown in Figure 3.14.
3.7. Conclusion
The target of this work is to observe the variation of jitter and packet end to
end delay in a WiMAX network under various mobility scenarios. This work
shows the variation of the mentioned parameters with respect to a stationary
network having all stationary nodes and with respect to a mobile network with
varying speed of 50km/hr, 100km/hr and 150km/hr. Average packet end to end
delay is more in case of stationary nodes than mobile nodes. Figures 3.7, 3.8 and
3.14 show the variation of jitter in the above mentioned networks. As Figure 3.7
and 3.8 shows, the jitter in a stationary network increases considerably when
silence suppression is activated in the network and it increases further when
mobility is introduced in the network. This is because as a node becomes mobile it
undergoes handover. This creates inconsistency in the order of the packets
delivered thereby jitter. Figures 3.9, 3.10 and 3.15 show the variation of packet end
to end delay in the above mentioned WiMAX network. From Figures 3.9 and 3.10,
it is observed that the packet end to end delay decreases in the overall network
when silence suppression is activated. This is due to the reduction in the queuing
delay. As the number of packets decreases with silence suppression, the number of
packets queued in the intermediate nodes decreases. This reduces the queuing
delay. The packet end to end delay decreases considerably when mobility is
introduced in the network. With mobility, the mobile nodes get distributed under
multiple base stations and this removes the bottle neck of one base station in case
of stationary nodes and thereby decreases the delay.
With a step towards the 4G mobile networks, this study is used in the
following chapters to compare and evaluate the performance of the loosely coupled
WiMAX WLAN integrated network.
38
Chapter 4
VoIP over WLAN Network
Chapter Outline: Introduction
 Background of IEEE 802.11 and WLAN
 Protocol Architecture
 Overview of IEEE 802.11 MAC Layer
 IEEE 802.11 MAC: DCF and PCF
 Overview of IEEE 802.11 PHY layer
 Frequency Hop Spread Spectrum (FHSS)
 Dynamic Sequence Spread Spectrum (DSSS)
 Infrared
 WLAN System Architecture
 Experimental Setup
 Scenario 1: Simulation Setup and Results
 Average Jitter
 Average Packet End to End Delay
 Average MOS
 Scenario 2: Simulation Setup and Results
 Average Jitter
 Average Packet End to End Delay
 Conclusion
39
VoIP over WLAN Network
4.1 Introduction
WLANs [1] are mostly designed for private wired LANs and have been
enormously successful for data traffic like (email, media downloads etc). WLANs
are being studied as an alternative to the high installation and maintenance costs
incurred by traditional addition, removal, and changes experienced in wired LAN
infrastructures. It is an economic way to provide users with ubiquitous connectivity
- anywhere, anytime - to the Internet and to private and corporate networks. It
offers typical gross data rate of 10-50 Mbps as opposed to 10-100 kbps offered by
cellular technology and compared to the peak 70 Mbps offered by WiMAX. Voice
traffic differs from data traffic mainly in its sensitivity to delay and loss [2].
Although existing WLAN applications are mainly data centric, there is a growing
demand for real-time voice services over WLAN. Driven by these two popular
technologies, VoIP over WLAN has been emerging as an infrastructure to provide
low-cost wireless voice services.
4.2 Background of IEEE 802.11 and WLAN
The first generation of WLANs operated in the 900 MHz ISM band, with
symbol rates of around 500 Kbps, but they were exclusively proprietary, nonstandard systems, developed to provide wireless connectivity for specific niche
markets. The second-generation systems came around 1997. They operated in the
2.4 GHz range and provided symbol rates of around 2Mbps. The standards of
WLAN are as follows:
o IEEE 802.11b (Wi-Fi 2.4 GHz): The goal of the Task Group b was to
increase the maximum bit rate in the 2.4 GHz frequency range while
maintaining interoperability with the original standard. The MAC layer was
kept and the PHY was redefine to only work with DSSS (Direct Sequence
Spread Spectrum), thus increasing the spectral efficiency with bit rates of
up to 11 Mbps.
40
o IEEE 802.11a (Wi-Fi 5.2 GHz): The goal of this group was to provide
higher data rates and to port IEEE 802.11 to the newly available U-NII at
5.2 GHz. The original MAC layer was kept and the PHY was reworked to
provide rates up to 54 Mbps. Since the available band at U-NII is about 300
MHz, eight non-overlapping bands were defined. The spread spectrum
technology used in this case was OFDM (Orthogonal Frequency Division
Multiplexing), as DSSS was not efficient at working with these high bit
rates.
o IEEE 802.11g: Task group g is working on an extension to IEEE 802.11b
at 2.4 GHz, enabling transmission at symbol rates of 54 Mbps.
There are also other IEEE 802.11 Task Groups that focus on different aspects such
as:
o 802.11i - wireless security at the Mac layer.
o 802.11f - roaming between access points
o 802.11e - the Quality of Service
The IEEE 802.11 WLAN standard that operate in the 2.4 GHz or 5 GHz
unlicensed radio bands; the four are summarized in Table 4.1.
TABLE 4.1. Characteristic feature of Wireless LAN
Standard Maximum Fallback
802.11b
Channels Frequency Radio
Bit Rate
Rates
Provided
Band
Technique
11 Mbps
5.5 Mbps, 2 Mbps
3
2.4 GHz
DSSS
5 GHz
OFDM
2.4 GHz
OFDM
1 Mbps
802.11a
54 Mbps
48 Mbps, 36 Mbps, 12
24 Mbps, 18 Mbps,
12 Mbps, 9 Mbps,
6 Mbps
802.11g
54 Mbps
Same as 802.11a
3
4.3 Protocol Architecture
Figure 4.1 shows the protocol Stack of IEEE 802.11. Part of the IEEE
802.11 protocol stack is shown in the figure 4.2. In this figure, only the relevant
parts of the protocol stack are depicted, as the upper layers are the same than the
41
other LANs and only the two layers represented (Physical Layer and Data Link
Control Layer) differ.
FIGURE 4.1. Protocol Stack
FIGURE 4.2. Part of the Protocol Stack
In the Physical Layer (PHY), several sub layers appear:
o Physical Medium Dependent (PMD). This sub layer’s functions are the
modulation and codification of the signal.
o Physical Layer Convergence Protocol (PLCP): This level is in charge of
adapting the PHY service to the PMD sub layer.
o Physical Management. This level selects the channel to transmit with.
Just above the Physical Layer, the Data Link Control Layer (DLC) is formed by
the following sub layers:
o Medium Access Control (MAC). This sub layer is responsible for the
access, fragmentation and ciphering of the signal.
o Logical Link Control (LLC). This level makes the assignment of the logical
channels to the physical channels.
o MAC Management. This level is responsible of the synchronization,
roaming, consume management, etc.
o Finally, the Station Management is responsible of the coordination of the
management functions in the two layers.
42
4.4 Overview of the IEEE 802.11 MAC
802.11 support two modes of operation. The first, called Distributed
Coordination Function (DCF), does not use any kind of central control and the
second, called Point Coordination Function (PCF), and uses the base station to
control all activity in its cell. So, DCF can be used by all implementations of
WLANs and PCF is optional. These two modes will be explained in detail.
4.4.1 DCF and PCF
The original IEEE 802.11 standard [3] specifies two channel access
mechanisms: a mandatory contention-based distributed coordination function
(DCF) and an optional polling based point coordination function (PCF). The DCF
uses a carrier sense multiple access with collision avoidance (CSMA/CA) scheme
for medium access and the optional four way handshaking request-to-send/clear-tosend [4] mechanism (RTS/CTS). DCF provides a best effort service and is not
capable of providing differentiation and prioritization based upon traffic type.
While DCF may provide satisfactory performance in delivering best-effort traffic,
it lacks the support for QoS requirements posed by real time traffic, and especially
VoIP which has stringent delay requirements. These requirements make the DCF
scheme a less feasible option to support QoS for VoIP traffic.
PCF mode, with a centralized controller, represented another promising
alternative to providing QoS in WLAN [5]. The centralized controller called the
point coordinator (PC) has its functionality embedded in the access point (AP) of a
BSS. The PC enables polling, enabling the polled stations to transmit data without
contending for the channel. Nevertheless, studies on carrying VoIP over WLAN in
PCF mode in [5] found that when the number of stations in a basic service set
(BSS) is large, the polling overhead is high. This results in excessive end-to-end
delay and VoIP gets poor performance under heavy load conditions.
Although the PCF mode is designed for real-time traffic, it is not widely
deployed due to its inefficient polling schemes, limited quality of service (QoS)
provisioning, and implementation complexity. Thus, both DCF and PCF have
limited support for real-time applications, the IEEE 802.11e has been proposed to
enhance the current 802.11 MAC to support applications with stringent QoS
requirements [7], but the wide deployment of the standard is yet to be done.
43
On the other hand, supporting voice traffic over WLANs using the DCF
mode poses significant challenges, because the performance characteristics of their
physical and MAC layers are much worse than their wired counterparts and hence
considered in our system.
4.5 Overview of IEEE 802.11 PHY layer
IEEE 802.11 supports three types of physical layer – Infrared, Frequency
Hop Spread Spectrum (FHSS), and Direct Sequence Spread Spectrum (DSSS). The
later two are based on radio transmission. All of them include the provision of the
clear channel assessment signal (CCA). The CCA is used for medium access and
indicates if the medium is idle. The transmission technology determines the
technique to obtain this signal. The PHY layer offers a service access point (SAP)
with 1 or 2 Mbps data rate to the MAC layer.
4.5.1 Frequency Hop Spread Spectrum (FHSS)
FHSS is a spread spectrum technique which allows for the coexistence of
multiple networks in the same area by separating different network using different
hopping sequence.
The standard specifies Gaussian shaped FSK (frequency shift keying),
GFSK, as modulation for FHSS PHY. For 1Mbps a two level GFSK is used (i.e. 1
bit mapped to 1 frequency) and for 2 Mbps a four level GFSK is used (i.e. 2 bits
are mapped to 1 frequency). Sending and receiving in 1Mbps is mandatory while
sending and receiving in 2 Mbps is optional. This facilitated the production of low
cost devices for the lower rate only and more powerful devices for both
transmission rates in the early days of 802.11.
A FHSS frame consists of two parts, namely the PLCP part (preamble and
header) and the payload part. PLCP part is always transmitted at 1Mbps while the
payload can use 1 or 2 Mbps. Also the MAC data is scrambled using the
polynomial s(z) = z7 + z4 + 1 for DC blocking and whitening of the spectrum.
4.5.2 Dynamic Sequence Spread Spectrum (DSSS)
Direct sequence spread spectrum is the alternative spread spectrum method
separating by code and not by frequency. In case of IEEE 802.11 DSSS, spreading
is achieved by using the 11-chip Barker sequence (+1,-1,+1,+1,-1,+1,+1,+1,-1,-1,1). The key characteristics of this method are its robustness against interference
44
and its insensitivity to multipath propagation (time delay spread). However, the
implementation is more complex compared to FHSS.
IEEE 802.11 DSSS PHY also uses the 2.4 GHz ISM band and offers both 1
and 2 Mbps data rates. The system uses differential binary phase shift keying
(DBPSK) for 1 Mbps and differential quadrature phase shift keying (DQPSK) for 2
Mbps as modulation schemes. The symbol rate is 1MHz resulting in a chipping
rate of 11 MHz all bits transmitted by the DSSS PHY are scrambled with the
polynomial s(z) = z7 + z4 + 1 for DC blocking and whitening of the spectrum.
4.5.3. Infrared
The PHY layer which is based on infra red (IR) transmission, uses near
visible light at 850-950nm. Infra red light is not regulated apart from safety
restrictions. The standard does not require a line of sight (LOS) between sender
and receiver, but should also work with diffuse light. This allows for point to
multipoint (PMP) communication. The maximum range is about 10m if no sunlight
or heat sources interfere with the transmission. Typically, such a network will only
work in buildings, e.g. classrooms, meeting rooms, etc. frequency reuse is very
simple. A simple wall is more than sufficient to shield one IR based IEEE 802.11
network from another. Today, no products are available that offer infra red
communication based on 802.11. Proprietary products offer, e.g. up to 4 Mbps
using diffused infra red light. Alternatively, directed infra red communication
based on IrDA can be used.
4.6 WLAN System Architecture
The architecture of a WLAN system is described according to the IEEE
802.11 standard. The main building block of such architecture is the Base Service
Set (BSS). There are two modes of configuration that can be used within the
standard:
o The “Infrastructure” and
o The “Ad Hoc” mode.
4.6.1 Infrastructure mode:
In the “Infrastructure” mode, the different BSS are interconnected with
each other via a component called the Distribution System (DS). Each BSS has one
Access Point (AP), through which the Mobile Nodes (MNs) access to the DS.
45
These interconnected components form the Extended Service Set (ESS). The ESS
is a large coverage area where MNs can get handed over from one BSS to another
without changes or notification to higher layers in the protocol stack. Finally, a
“Portal” is required to integrate the WLAN architecture into the wired network
(e.g., Ethernet) and may be integrated with an AP in a single device attached
directly to the DS. Fig 4.3 shows the WLAN topology in infrastructure mode.
FIGURE 4.3. Infrastructure based architecture
4.6.2 Ad-Hoc mode:
In the “Ad-Hoc” mode, each MN can directly reach any MN within the
BSS without going through an intermediate node (i.e., AP). There is no backbone
network or distributed system associated with the mobile nodes and the BSS are
not connected to the wired network. The Adhoc mode of connection usually covers
small area and the nodes are allowed to move only within their BSS. There is no
concept of handover to support mobility across a large area. Fig 4.4 shows the
topology of WLAN in Adhoc Mode.
FIGURE 4.4. Adhoc Architecture
46
4.7 Experimental Setup
To analyze VoIP in a network, it is necessary to study real life scenarios.
Hence, OPNET 14.5.A is chosen as the simulation tool as it will reflect the actual
deployment of the WiMAX network.
4.7.1 Scenario 1 – Static Nodes:
Figure 4.5 shows the simulation setup used for WLAN (IEEE 802.11b)
network. Using the Wireless Deployment Wizard of OPNET a 7 celled WLAN
network, with multiple subscriber stations in the range of an Access Point (AP) is
deployed. The AP is connected to the core network by a server backbone via an IP
backbone. The server backbone is further connected to the voice server which is
configured as the SIP server. The base station, IP Backbone, Server Backbone and
the Voice Server together represent the service provider company network. The
cell radius is set to be 100 m. The numbers of subscribers in cells are 10. Voice
over IP calls are setup between the subscriber stations of cells 2 and 3 in mesh
using SIP. The parameters set in the Access Points and the Subscriber Stations are
shown in figures 4.6 and 4.7 respectively.
FIGURE 4.5. Network Model for WLAN in OPNET
FIGURE 4.6. Access Point Parameters
47
FIGURE 4.7. Subscriber Station Parameters
With the simulation setup as mentioned, the common voice codecs being
used for the Voice over IP calls are varied and the corresponding variation in voice
jitter, MOS (Mean Opinion Score) and Packet End-to-end delay are noted.
4.7.1.1 Average Jitter
Figure 4.8 shows that all voice codecs except G 723.1 has non zero voice
jitter with positive variation. Voice quality is considered to be best if the jitter
value is zero. Thus, it can be said from this graph that G 723.1 provides best
performance with zero value and G 729 on the other hand produce worst
performance with highest jitter of 0.001 seconds. But as stated in [5], since G
723.1 is a low bit rate codec, it generates small packets of 5.3 Kbps and 6.3 Kbps
and hence can only be used for narrow band communications. Other than G 723.1,
G 711 shows lower jitter compared to others at the cost of bandwidth consumption.
G 711 is a high bit rate codec and uses 64Kbps bit rate for data transfer. It provides
PSTN quality of voice. As already mentioned in chapter 1 voice communication
has 35% voice and 65% silence in it. To accommodate more users, silence present
in the voice communication needs to be suppressed. This is expected to save
bandwidth. Results pertaining to silence suppression are shown in figure 4.9.
Figure 4.9 shows that the order of voice jitter decreases to an order of 10-5 from 103
when silence is suppressed. This is quite less. After silence suppression G 711
shows highest jitter among all the codecs considered. G 726 (32Kbps) shows
average position in both the graphs indicating optimal performance. According to
[7], G 726 has its roots in the PSTN network and provides considerably good
quality of voice. Since, it shows good performance in both the cases (figures 4.7
and 4.8) it can be used in trunks and thereby can accommodate more users by
saving bandwidth.
48
Average Jitter without using silence suppression
0.0012
Second(sec)
0.001
0.0008
0.0006
0.0004
0.0002
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Codecs
FIGURE 4.8. Average voice jitters without Silence Suppression
Average Jitter using silence suppression
0.000016
0.000014
Seconds(sec)
0.000012
0.00001
0.000008
0.000006
0.000004
0.000002
0
-0.000002
G729
G711
G723.1 G723.1
5.3K
6.3K
G726
G726
16K
24K
Voice Code cs
G726
32K
G726
40K
G728
12.8K
G728
16K
FIGURE 4.9. Average voice jitters using Silence Suppression
4.7.1.2 Average Packet End to End Delay
The average packet end to end delay in WLAN network for voice codecs
without silence suppression is shown in figure 4.10 and is in the order of seconds.
This is because WLAN 802.11b provides a bandwidth of 11 Mbps and the Access
Points 2 and 3 [refer to figure 4.5] are overloaded and congested with the voice
traffic. This results in a bottleneck situation in the Access Points which results in
packet drop due to queue overflow in the access points. Any packet loss in the
physical and MAC layers is interpreted as delay by the higher layers. Hence, the
delay. Except G 723.1, all other voice codecs shows an average delay of more than
2 seconds if silence period is not suppressed. G 723.1 gives a packet end to end
delay of less than 0.25 seconds which is acceptable but can only be used for
narrow band communication. With silence suppression the packet end to end delay
is considerably less and in the order of 10-2 seconds. The result is shown in figure
4.11. G 723.1 though giving a good performance without silence suppression
shows terrible result when silence is suppressed. The packet end to end delay for G
723.1 increases up to 0.1 seconds while the same for other is about 0.06
49
approximately which is quite less comparatively. G 723.1 creates small packets of
larger count. Though silence suppression decreases its number of packets but still
its packet count remains quite high compared to the other codecs. This results in
network congestion and thereby increased delay. Also, the look ahead delay of G
723.1 is 7.5msec [8] while the same for G 729 is 5sec and others are zero [8]. This
adds up to the delay of G 723.1.
Average Packet End to End Delay without using silence suppression
5
Delay(sec)
4
3
2
1
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726 16K G726 24K G726 32K G726 40K
G728
12.8K
G728 16K
Voice Codecs
FIGURE 4.10. Average packet end to end delay without Silence Suppression
Average Packet End to End Delay using Silence Suppression
0.12
Delay (sec)
0.1
0.08
0.06
0.04
0.02
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice codecs
FIGURE 4.11. Average packet end to end delay using Silence Suppression
4.7.1.3 Average MOS
Figure 4.12 shows the Mean Opinion Score (MOS) obtained by the voice
codecs in a WLAN network without using silence suppression. It shows that the
voice quality produced by all voice codecs except G 723.1 is unacceptable as they
have scored a MOS value of 1 which implies unacceptable quality of voice [5].
Hence, silence suppression has to be introduced and the improved results due to
silence suppression are shown in figure 4.13. With introduction of silence
suppression, the MOS value shows a considerable rise for all voice codecs except
G 723.1. For most of the voice codecs the MOS value has exceeded 3.5 indicating
very good quality of voice. Hence to finally select a voice codec in a static WLAN
50
network accommodating large number of users, the voice codec G 726 is expected
to perform the best.
Average MOS without using silence suppression
Average MOS
3
2.5
2
1.5
1
0.5
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Codecs
FIGURE 4.11. Average MOS without Silence Suppression
Average MOS
Average MOS using silence suppression
4
3.5
3
2.5
2
1.5
1
0.5
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice codecs
FIGURE 4.12. Average MOS using Silence Suppression
4.7.2 Scenario 2 – Mobile Nodes:
Figure 4.13 shows the simulation setup used for roaming enabled WLAN
network. Using the Wireless Deployment Wizard of OPNET a 7 celled WLAN
network, with multiple subscriber stations in the range of a base station is
deployed. The access points are connected to the core network by a server
backbone via an IP backbone. The server backbone is further connected to the
voice server which is configured as the SIP server. The base station, IP Backbone,
Server Backbone and the Voice Server together represent the service provider
company network. The cell radius is set to be 100 m. The mobile nodes are
configured to move at a rate of 50 km/hr, 100 km/hr and 150 km/hr along the path
shown in white lines. The number of subscribers in all cells is 10. Voice over IP
calls are setup between nodes of cells 2, 3 and 7 using SIP and are shown by the
blue lines in the figure 4.13.
51
FIGURE 4.13. Network Model for WLAN
4.7.2.1 Average Jitter
Figure 4.14 shows the variation of jitter across the mobility supported
WLAN network with respect to the various voice codecs considered. From the
figure it is observed that the average jitter irrespective of speed decreases with the
decrease in data rate. G 711 has the highest data rate of 64Kbps and has highest
jitter. G 726 has a data rate of 32Kbps as considered and shows jitter less than G
711. G 728 and G 729 has data rates of 16Kbps and 8Kbps and thereby shows the
decrease in the jitter and finally G 723.1 which has a minimum data rate of
5.3Kbps and has minimum jitter. As the data rate decreases the load of the network
decreases. As the load decreases, the router processing happens in time and the
packets arrive in time. This thereby decreases the jitter.
Average Jitter
0.35
0.3
Seconds
0.25
50km/hr
0.2
100km/hr
0.15
150km/hr
0.1
0.05
0
G 711
G 723
G 726
G 728
G729
Voice Codecs
FIGURE 4.14. Average Jitter
4.7.2.2 Average Packet End to End Delay
As mobility is introduced in the network, the network performance
degrades to an unacceptable level. As the mobile nodes start moving the packet
end-to-end delay increases in the order of 10 seconds. This shows that due to low
bandwidth a WLAN network is unable to support voice communication at
vehicular speed. Overall we can say that such huge delay makes WLAN incapable
of supporting voice communication.
52
Seconds
Average Packet End to End Delay
180
160
140
120
100
80
60
40
20
0
50km/hr
100km/hr
150km/hr
G 711
G 723
G 726
G 728
G729
Voice Code cs
Fig 4.15 Average Packet end to end delay
4.8 Conclusion
The target of this work is to observe the variation of jitter and packet endto-end delay in a WLAN network under various mobility scenarios. This work
shows the variation of the mentioned parameters with respect to a stationary
network having all stationary nodes and with respect to a mobile network with
varying speed of 50km/hr, 100km/hr and 150km/hr. Unlike the WiMAX network
average packet end to end delay is less in case of stationary nodes than mobile
nodes. Figures 4.8, 4.9 and 4.14 show the variation of jitter in the above mentioned
networks. As figure 4.8 and 4.9 shows, the jitter in a stationary network decreases
considerably when silence suppression is activated in the network and it increases
quite a lot when mobility is introduced in the network. This is because as a node
becomes mobile it undergoes handover. This creates inconsistency in the order of
the packets delivered thereby jitter. Figures 4.10, 4.11 and 4.15 show the variation
of packet end to end delay in the above mentioned WLAN network. From figures
4.10 and 4.11, it is observed that the packet end to end delay decreases in the
overall network when silence suppression is activated. This is due to the reduction
in the queuing delay. As the number of packets decreases with silence suppression,
the number of packets queued in the intermediate nodes decreases. This reduces
the queuing delay. The packet end to end delay increases largely when mobility is
introduced in the network. With mobility, the mobile nodes move at a very high
speed over the network and while getting handed over from one AP to another it
moves to a third AP. This results in the fact that the speed of handover is less
compared to the speed of the mobile node. Hence, the network fails and the effect
is visible to us in terms of huge delay.
With a step towards the 4G mobile networks, this study is used in the
following chapters to compare and evaluate the performance of the loosely coupled
WiMAX WLAN integrated network.
53
Chapter 5
VoIP over WLAN-WiMAX Integrated
Network
Chapter Outline: Introduction
 WLAN vs. WiMAX
 Types of Integration
 Loose Coupling
 Tight Coupling
 Working of the Integrated Network
 Scenario 1: Stationary Nodes
 Scenario 2: Mobile Nodes
 Experimental Setup
 Scenario 1: Simulation Setup and Results
 Average Jitter
 Average Packet End-to-End Delay
 Average MOS
 Scenario 2: Simulation Setup and results
 Average jitter
 Average Packet End-to-End Delay
 Conclusion
54
VoIP over WLAN-WiMAX Integrated
Network
5.1 Introduction
With growing demand for anytime anywhere connectivity, wireless
networks are gaining popularity and have become an integral part of human life.
Researchers of academia and industry are searching for new innovative methods to
cater to the requirement of the variety of roaming users. Nowadays the requirement
is not only restricted to connectivity but quality of service is also added to the
requirement. With voice over IP coming into existence this demand has gained
more importance. By distributing high-speed internet access from cable, digital
subscriber line (DSL), and other fixed broadband connections within wireless
hotspots, WLAN has dramatically increased productivity and convenience. The
integration of WLAN into laptops has accelerated the adoption of WLAN and an
increased number of handhelds and Consumer Electronics (CE) devices are
thereby adding WLAN capabilities. Moreover as shown in Chapter 4, the quality
of service for voice application is of acceptable quality in WLAN network.
WiMAX on the other hand has taken wireless Internet access to the next
level, and over time, could achieve similar attach rates to devices as WLAN.
WiMAX can deliver Internet access miles from the nearest WLAN hotspot and
blanket large areas -Wide Area Networks (WANs), be they metropolitan,
suburban, or rural - with multi-megabit per second mobile broadband Internet
access. In the last few years, WiMAX has established its relevance as an alternative
to wired DSL and cable, providing a competitive broadband service offering that
can be rapidly and cost effectively deployed. Now, Mobile WiMAX, as defined in
the Institute of Electrical and Electronic Engineers (IEEE) 802.16e-2005 standard,
adds broadband connectivity on the move. Mobile WiMAX, based on scalable
Orthogonal Frequency Division Multiple Access (OFDMA) technology, is capable
of simultaneously supporting fixed, portable, and mobile usage models.
Together, WiMAX and WLAN are ideal partners for service providers to
deliver convenient, affordable mobile broadband Internet services in more places.
55
Both are open IEEE wireless standards built from the ground up for Internet
Protocol (IP)-based applications and services. By using WiMAX backbone and
WLAN hotspot, service providers can deliver high-speed Internet connectivity that
subscriber’s desire in more places. And WiMAX and WLAN technology synergies
enable seamless integration into laptops, CE devices, and generate a new category
of devices called “mobile Internet devices”. These devices are expected to have
dual stack for both WLAN and WiMAX.
5.2 WLAN vs. WiMAX
Although both WiMAX and WLAN provide wireless broadband
connectivity, they have been optimized for different usage models: WLAN for
very high-speed WLAN connectivity and WiMAX for high-speed Wireless WAN
(WWAN) connectivity. By combining WiMAX and WLAN technologies, service
providers can offer their subscribers a more complete suite of broadband services
in more places. Table 5.1 illustrates how WiMAX and WLAN complement each
other from an implementation and deployment perspective and also shows the
impact of the WLAN WiMAX integration synergy.
TABLE 5.1: WLAN and WiMAX Comparison [42] [43]
WiMAX 802.16e
WLAN
Synergy Impact
802.11b
Primary
Deployed in wide Deployed
Application
coverage
in “Best-connected”
areas, local coverage model: users connect
areas, such as to WiMAX or WLAN
including
metropolitan areas public
depending on their
for
location,
mobile hotspots,
broadband
homes
wireless as well as businesses.
rural
or
coverage,
and and Quality of Service
(QoS) requirements.
remote
areas for last-mile
connectivity
and
portable service
Frequency
Licensed/
Band
Unlicensed
2.4 GHz ISM Service providers can
2-11 Band
leverage both types of
56
GHz
spectrum;
for
example,
license
exempt for best effort
local area traffic and
licensed for wide area
and
QoS
sensitive
traffic.
Coverage
Metropolitan area Short
range Full range of services
mobile coverage of with up to 300 in
up
to
the
home
and
several meters for a office, as well as on
kilometers for a single
access the road. Economical
single base station. point.
coverage
Longer range (up
areas;
to several miles)
WLAN
for fixed & lower-
cafés,
density
airports, and WiMAX
deployments
for blanket coverage
of
for
large
example,
hotspots
hotels,
in
and
outside of hotspots.
Voice
over Both
over VoIP is supported Voice
IP
specifications
by the extended Internet
support
real-time
however, operations in
polling Protocol
class of service.
(VoIP) is
VoIP;
license-exempt
supported with spectrum limit QoS
enhancements
Mobility
assurance.
Provides fixed and Provides fixed Full range of services
portable solutions and
portable in
over a wide area of solutions
1-3 miles.
within
the
home
and
office, as well as on
the the road.
coverage of the
hot spot.
Deployment
Customer Premise Embedded
in Integration
into
Equipment (CPE) 97% of laptops devices is expected to
and PC cards will and many hand reduce
device
57
be
embedded in held and CE subsidies and lower
laptops and hand devices.
Cost Per Gross Add
held devices
(CPGA)
5.3 Types of integration
Mobility across WLAN and WiMAX can be integrated according to two
models namely loose coupling and tight coupling. Since, both the networks are IP
based; integration does not require translations in protocols when bridging the
infrastructure.
5.3.1 Loose Coupling
The lack of network management features in WLAN requires additional
components by the WLAN operator to obtain a higher degree of network
intelligence when performing vertical handover. Network intelligence can be used
to facilitate context transfer of AAA and services prior to handover. This will
reduce the latency of the handover and make the transition seamless. Loose
coupling enables flexibility for operators as roaming agreements can be made to
extend WiMAX data coverage without expanding or integrating WLAN in a
WiMAX network. Crucial to global roaming between WLAN hot spots and
WiMAX networks is better service tracking in WLAN where accounting and
billing is supported as WLAN currently lacks this facility.
5.3.2 Tight Coupling
In this type of integration, components of WiMAX network are already
present to support user mobility and AAA and integration proxies can be inserted
in the core network to support seamless connection transfers while avoiding dual
traffic flows. Care needs to be taken with such a solution so that end-to-end
semantics are preserved. Proxies also add a point of failure which requires some
redundancy, but managing at a higher level in the core infrastructure could mean
better scalability. Location management would also be made easier with the
knowledge of network infrastructure when moving between networks. With
location management, mobile devices can refrain from searching for other network
topologies on other interfaces, thus saving power. A WLAN attached to a WiMAX
58
network could use the CSN to supply users with IP-connectivity and IP-address
allocation.
5.4 Experimental Setup
Loosely coupled system has been developed and its performance with
respect to voice over IP is checked for varying voice codecs. The simulation setup
is as shown in the following sections.
5.4.1 Scenario 1: Stationary nodes
An integrated scenario comprising of WLAN hot spots and WiMAX
backbone is developed. The WiMAX backbone comprises of two base stations
which are connected to the SIP server via the routers. A special type of node called
SS_WiMAX_WLAN_AP having dual stack and behaving both as WLAN AP and
WiMAX subscriber station is placed under the coverage of the Base Station. This
node at the same time behaves both as the Subscriber Station of the WiMAX
network and Access Point of the WLAN hot spot and acts like a bridge. There are
10 WLAN workstations under each such Access Points and VoIP calls are setup
between the workstations of the two APs. The parameters of the WiMAX Base
Station and the WiMAX interface of SS_WiMAX_WLAN_AP are same as that of
WiMAX BS and WiMAX SS set in Chapter 3 and the parameters of the WLAN
workstations and WLAN interface of the SS_WiMAX_WLAN_AP are same as
shown in Chapter 4. The scenario is as shown in Figure 5.1 and the protocol stack
of SS_WiMAX_WLAN_AP is shown in Figure 5.2. The entire process of
developing the scenario is discussed in Chapter 6. With the simulation setup as
mentioned, the common voice codecs being used for the Voice over IP calls are
varied and the corresponding variation in voice jitter, MOS (Mean Opinion Score)
and Packet End-to-end delay are noted.
FIGURE 5.1 Network Model for WiMAX-WLAN Static Integrated Network
59
FIGURE 5.2. Protocol Stack of SS_WiMAX_WLAN_AP
5.4.1.1 Average Jitter
Figure 5.3 shows the variation of jitter for all voice codecs. It is seen that
the voice codec G 723.1 exhibits almost zero jitter which implies very good quality
of voice while others exhibit jitter varying from 0.0004 to 0.0005 seconds and
about 0.00055 for G729. The result when silence period is suppressed which is
about 65% of th entire communication period and is shown in Figure 5.4.
Average Jitter
0.0006
0.0005
Seconds
0.0004
0.0003
Integrated
0.0002
0.0001
16
K
G
72
8
G
71
G
1
72
3.
1
5.
3K
G
72
3.
1
6.
3
K
G
72
6
16
K
G
72
6
24
K
G
72
6
32
K
G
72
6
40
K
G
72
8
12
.8
K
G
72
9
0
-0.0001
Voice Code cs
FIGURE 5.3 Average voice Jitter without Silence Suppression
Figure 5.4 shows the variation of jitter in the voice over IP for varying
voice codecs with voice activity detection enabled. It detects the silence period and
prevents paketisation of the background noise during the silence period. It is
observed from the graph that average voice jitter decrease to the order of 10-5 to 104
which is very nominal. Interestingly, G 723.1 shows unchanged performance.
Voice Activity Detection and silence suppression does not affect the performance
of G 723.1 with respect to jitter. This is due to the fact that, G 723.1 is a low bit
rate codec and hence generates small packets. This thereby has much decreased
load even without silence suppression on the network unlike other codecs. Low
load results in low processing time and increased network performance. Also
60
among the two data rates of G 723.1, it is seen in figure 5.4 that the lower data rate
generates lower jitter than the higher data rate codec. G 711 having the highest data
rate displays maximum jitter.
Av e rage Jitte r
1.20E-04
1.00E-04
8.00E-05
Seconds
6.00E-05
4.00E-05
Integrated
2.00E-05
16
K
12
.8
K
G
72
8
40
K
G
72
8
G
72
6
32
K
24
K
G
72
6
16
K
G
72
6
5.
3
K
6.
3
K
G
72
6
-4.00E-05
G
72
3.
1
G
71
1
-2.00E-05
G
72
3.
1
G
72
9
0.00E+00
-6.00E-05
Voice Code cs
FIGURE 5.4 Average voice Jitter with Silence Suppression
5.4.1.2 Average MOS
With MOS coming into the picture, figure 5.5 shows the mean optimal
score (MOS) scored by the various voice codecs in the WLAN-WiMAX integrated
network. As shown in the figure the highest MOS obtained by the voice codecs
with background noise packetised during the silence period i.e. without silence
suppression reaches a maximum value of 2.5 and for G 723.1 while for others it
ranges from 1 to 2. From the MOS value we can conclude that the network is
incapable of supporting VoIP traffic.
A totally contradicting picture is observed in the figure 5.6 which shows the
performance of voice codecs with silence being suppressed in the communication
period. The MOS value scored jumps from the range of 1-2 to 3-3.5 implying the
network can provide acceptable quality of voice. Here comes the importance of
silence suppression. Again like average jitter the average MOS for the voice codec
G 723.1 remains unchanged irrespective of silence suppression for the same reason
as stated for jitter.
Ave rage MOS
3
2.5
MOS
2
1.5
1
0.5
16
K
G
72
8
12
.8
K
G
72
8
32
K
40
K
G
72
6
G
72
6
24
K
G
72
6
16
K
G
72
6
K
6.
3
5.
3
K
G
72
3.
1
G
72
3.
1
G
71
1
G
72
9
0
Voice Code c
FIGURE 5.5 Average voice MOS without Silence Suppression
61
Ave rage M OS w ith Silence Suppr es s ion
16
K
G
72
8
40
K
32
K
24
K
12
.8
K
G
72
8
G
72
6
G
72
6
G
72
6
G
72
6
6.
3
16
K
K
K
5.
3
G
72
3.
1
G
71
1
G
72
3.
1
G
72
9
MOS
4
3.5
3
2.5
2
1.5
1
0.5
0
Voice Code cs
FIGURE 5.6 Average voice MOS with Silence Suppression
5.4.1.3 Average Packet end-to-end delay
The packet end-to-end delay for the WLAN-WiMAX integrated network is
shown in figure 5.7. As shown in the figure, the packet end-to-end delay for all
voice codecs except G 723.1 is in the order of seconds varying from 1.5 sec for G
711 to above 3 seconds for G 728 and G 729. G 723.1 exhibits packet end-to-end
delay of less than 0.5 seconds. This is because G723.1 is a low bit rate codec and
generates small packets at the rate of 5.3Kbps or 6.3 Kbps compared to others.
Packet End-to-End Delay
4
3.5
Seconds
3
2.5
2
1.5
1
0.5
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Codecs
FIGURE 5.7 Average Packet end-to-end delay without Silence Suppression
Pa cket End-to-End Delay with Silence Suppression
0.14
0.12
Seconds
0.1
0.08
0.06
0.04
0.02
0
G729
G711
G723.1
5.3K
G723.1
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Code cs
FIGURE 5.8 Average Packet end-to-end delay with Silence Suppression
The overall delay decreases to the order of 10-2 seconds when silence
suppression is introduced and is shown in figure 5.8. Decreased delay is due to the
decreased load of the network. Silence suppression basically identifies the silence
62
period in the communication period by sensing the amplitude and stops packetising
the voice when silence is encountered. This decreases the number of packets in the
network thereby releasing the network load.
5.4.2 Scenario 2: Mobile Nodes
A scenario having two WLAN hot spots and a large span of area between
the hot spots covered by WiMAX is setup. The scenario is shown of Figure 5.3.
The WiMAX part of the network is deployed by the Wireless deployment wizard
of OPNET and WLAN Access Points are taken from the object palette. 10 special
nodes are configured to move from one Access point to the other access point i.e.
from one hot spot to another hot spot while communicating with 5 nodes of cell
number 3 of WiMAX and 5 nodes of cell number 7. Results show that the mobile
nodes are supported and traffic is send to them even when they are under the
jurisdiction of WiMAX network. The steps to develop the scenario are discussed in
Chapter 6.
FIGURE 5.9 Network Model for WiMAX-WLAN Integrated Mobile Network
5.4.2.1 Average Jitter
The average voice jitter is found to be in the order of 10-2 which is quite
high than static scenario. This is because as the mobility is introduced the routing
path varies. This results in a non serial way of packet arrival. Some packets reach
before time while others reach with a time lag resulting in increased jitter. Also it
63
is observed that jitter increases with speed. As the speed of the mobile node
increases, the mobile node gets handed over faster from the jurisdiction of one
BS/AP to another. This fast change of routing path causes frequent routing update
which in turn results in unordered packet and hence increased jitter.
Average Jitter
0.03
Seconds
0.025
0.02
50km/hr
0.015
100km/hr
0.01
150km/hr
0.005
0
G711
G723
G726
G728
G729
Voice Codecs
FIGURE 5.10 Average Voice Jitter without silence suppression
5.4.2.2 Average Packet end-to-end delay
The mobile scenario shows a decreased packet end-to-end delay. This is
shown in figure 5.11. As the mobile node starts moving, the load on the WLAN
APs which creates a bottleneck scenario is released. The entire load gets
distributed among the WiMAX base stations. This results in a fall in the packet
end-to-end delay. Also as the load of the packets decreases, the processing time of
the BS and the APs decreases considerably thereby saving time. Delay for G 711 is
observed to be quite high compared to the other voice codecs. This is because; G
711 is a high bit rate codec and the rate of packets in about 64 Kbps. This results in
a huge processing time by the router hence increased delay. The delay increases
with speed. As the speed increases the mobile node faces increased number of
hand over and hence the routing path has to be recalculated which is time
consuming. Also the packet reassembly takes time and adds up to the delay.
Packet End-to-End Delay
0.3
Seconds
0.25
0.2
50km/hr
0.15
100km/hr
0.1
150km/hr
0.05
0
G711
G723
G726
G728
G729
Voice Codecs
FIGURE 5.11 Average Packet end-to-end delay without silence suppression
64
5.5 Conclusion
Overall it is concluded that, silence suppression improves the performance
of the network with respect to jitter, MOS and Packet end to end delay due to the
decrease in the network load. With mobility being introduced in the network the
average jitter increases considerably but the packet end-to-end delay decreases
remarkably. Due to the limitation of OPNET, the MOS value for the mobile
scenario could not be obtained but is since MOS is dependent on the delay of the
network, it is expected that the MOS value will increase in the mobile scenario.
Also, the network performance degrades with increase in speed.
65
Chapter 6
Development of Scenarios in OPNET
Chapter Outline:-
 Why OPNET?
 Development of WiMAX in OPNET
 Static Scenario
 Mobile Scenario
 Development of WLAN in OPNET
 Static Scenario
 Mobile Scenario
 Development of the WLAN-WiMAX Integrated Scenario
 Static Scenario
 Mobile Scenario
66
Development of Scenarios in OPNET
6.1 Why OPNET?
OPNET
comprehensive
(Optimized
development
Network
Engineering
environment
Tool)
supporting
the
[44]
provides
modeling
of
communication networks and distributed systems. Both behavior and performance
of the modeled systems can be analyzed by discrete event simulation. Tool for all
phases of our study including model design, simulation, data collection and data
analysis are incorporated in OPNET environment. Various constructs pertaining to
communication and information processing are provided by OPNET. Thus it
provides high leverage for modeling and distributed systems. Graphical
specifications of a model are provided by OPNET most of the times. It provides a
graphical editor to enter the network and model details. These editors provide an
intuitive mapping from the modeled system to the OPNET model specification.
OPNET provides 4 such type of editors namely the network editor, the node editor,
the process editor and parameterized editor organized in a hierarchical way. It
supports model level reuse i.e. models developed at one layer can be used by
another model at a higher layer. All OPNET simulations automatically include
support for analysis by a sophisticated interactive debugger. Technology
developers leverage advanced simulation capabilities and rich protocol model
suites to design and optimize proprietary wireless protocols.
In this work we have taken the advantages of OPNET Wireless modeler
suites (OPNET 14.5.A) particularly the WiMAX and WLAN platform for our
performance study.
6.2 Development of WiMAX in OPNET
To deploy a large 7 cell WiMAX network and examine the performance of voice
applications in static and mobile conditions the following steps are to be followed:
Methodology
1. Create a 7 cell WiMAX network and connect it to an IP backbone
2. Deploy voice applications between WiMAX nodes and configure the voice
server as the SIP Server.
67
3. Deploy voice QoS for WiMAX by choosing and tailoring a service class.
Map voice applications to WiMAX node by creating classifiers and service
flows on WiMAX nodes
4. Examine the performance of voice application in static conditions
5. Deploy trajectories on WiMAX SS nodes to make them mobile. Use ASN
anchored mobility for handover.
6. Deploy Application Demands.
7. The impact of mobility on application performance is then observed.
6.2.1 Static Scenario:
Steps to develop the static scenario are as follows:
1. Start OPNET Modeler by double-clicking on the OPNET Modeler 14.5
icon.
2. Select File > New.
3. A new window appears. Select Project from the menu and click OK
4. Another window appears and asks for project name and scenario name.
Enter project name as WiMAX_OPNET and scenario name as Static.
Uncheck the check box to not use the startup wizard and click OK.
5. Select Topology > Deploy Wireless Network. This launches the Wireless
Network Deployment Wizard.
FIGURE 6.1. Wireless Network Deployment Wizard
6. Click Continue and next until the Technology tab is reached. Select the
technology as WiMAX and set the Subscriber node transmission power as
0.5 W and Base Station transmission power as 10 W. Click next and move
to the next Tab of Topology.
68
FIGURE 6.2. Technology Window
7. In the topology tab enter the cell radius to be 30 km and from the drop
down menu select the node placement to be circular. This is because
circular nodes are located equidistantly from the base stations. The settings
are shown in the following figure.
FIGURE 6.3. Topology Window, First Tab
8. On clicking next the second tab of topology appears. Select the Subscriber
Station as wimax_ss_wkstn_adv from the dropdown list. This node allows
SIP configuration. Click Next.
9. The node mobility tab appears. Since we are deploying a static network
click on the row and click on delete row. Click Next.
10. The configuration summary appears as follows.
69
FIGURE 6.4. Configuration Summary
11. Click Finish. The deployed scenario consists of 7 celled WiMAX network
with the 7 BSs connected to a backbone.
12. To select Free Space as the pathloss model, Right-click on any of the
mobile nodes, and then choose Select Similar Nodes. Right-click on any
node again, and select Edit Attributes on any selected node in the scenario
space.
13. Check Apply changes to selected objects.
14. Expand
WiMAX
Parameters/SS
Parameters
and
set
Pathloss
Parameters to Free Space.
15. Click OK needed to apply changes to all selected nodes.
16. Click Yes to dismiss the compound attribute warning. You should see the
message “35 objects changed” in the status bar.
FIGURE 6.5. Subscriber Station Parameters
70
17. Click on the palette icon (
), scroll up to the ppp_server and drag and
drop it on the work space. Similarly, drag an ip32_cloud and drop it on the
workspace. Rename the ppp_server as the Voice_Server.
18. Now scroll up to the Links_PPP and drag a PPP_SONET_OC1 link to
connect
the
ppp_server
and
the
IP
cloud.
Similarly
select
a
PPP_SONET_OC_12 link and connect the IP cloud to the node called
backbone. The scenario now looks as follows. Right-click in the project
workspace after connecting the link to deactivate the handler.
FIGURE 6.6. Deployed WiMAX Scenario
19. Close the Object Palette.
20. From the object palette bring a Application Config and a Profile Config
node to configure the application.
21. Right click on the Application Config and select Edit Attributes
(advanced).
22. Under Application Definition add a row.
23. Enter Application Name as voice.
24. Expand the Description menu select the value of voice menu to edit.
25. Select the encoding scheme.
26. The voice codecs with (silence) implies silence suppression. By default
OPNET takes 65% of the communication period as silence and 35% as the
talk spurt.
27. The Profile Config node is configured as shown in the following figure.
71
FIGURE 6.7. Profile Setting
28. Now with the application and profile configured, we need to deploy the
application.
29. Select Protocols > Applications > Deploy Defined Applications
30. On the right-hand pane, expand Profile:”voice_profile” > Source and
click on Source. On the left-hand pane, while pressing Ctrl, select all
nodes of Cell 2. Once all of these nodes are selected, click on the
assignment arrow (>>) between the two panes.
31. Similarly, on the right-hand pane, expand Application:” voice” > Tier:”
Voice Destination” and click on “Voice Destination.” On the left-hand
pane, select all nodes of cell 3, and click on the assignment arrow between
the two panes. At this point, your right-hand pane should look like the
following figure. Click OK.
FIGURE 6.8. Application Deployment
72
32. The deployed voice application will be given strong QoS guarantees in
WiMAX network. For this reason, we will choose a service class using
ertPS scheduling.
33. Go to View/Show Network Browser; in the network browser pane, rightclick to Edit Attributes on the WiMAX_Config node. This node contains
the global parameters of the WiMAX network and is deployed
automatically when a WiMAX network is deployed. Edit the MAC Service
Class Definitions attribute by double-clicking in the corresponding value
field.
34. In the first row, modify the Gold service class by setting:
1. Scheduling Type to ertPS
2. Maximum Sustained Traffic Rate (bps) to 96000
3. Minimum Reserved Traffic Rate (bps) to 96000
4. Maximum Latency(milliseconds) to 10
35. Click OK once to accept changes.
36. On the WiMAX_Config node, set the Efficiency Mode to Physical Layer
Enabled.
37. Then click OK to accept the changes.
38. Right-click on any Mobile node in the network browser pane. The wireless
deployment wizard always deploys the WiMAX_Config node in the top
subnet, which explains its position outside the current subnet.
39. Deploy service flows and classifiers on the WiMAX mobile nodes, as
follows:
1. For each node of Cell 2, on which the voice application has been
deployed, deploy ertPS connections for both uplink and downlink
directions:
2. Select these nodes in the network browser (by holding down the Ctrl
key while selecting)
3. Right-click on one of these selected nodes in the scenario space, choose
Edit Attributes and checkmark Apply changes to selected objects.
4. Expand WiMAX Parameters/SS Parameters and click on the Value
field of Downlink Service Flows; add one row in the Downlink
Service Flows table and set:
1. Service Class Name to Gold
73
2. Modulation and Coding to QPSK 1/2; click OK once to accept
changes.
5. Similarly, add one row to the Uplink Service Flows table and set the
first row as follows:
1. Service Class Name to Gold
2. Modulation and Coding to QPSK 1/2; click OK once to accept
changes.
6. While still editing the attributes of the mobile node, add a classifier
definition that directs voice traffic into the newly configured ertPS
connections for uplink traffic.
1. Click on the Value field of WiMAX Parameters/Classifier
Definitions and add one row to the Classifier Definitions table; in
the new row set:
1. Type of SAP to IP
2. Traffic Characteristics to IP ToS, Equals, Interactive Voice
(6)
3. Service Class Name to Gold
4. Click OK once to accept classifier changes.
7. Click OK one final time to apply these changes to the selected objects.
8. Click Yes to ignore compound attribute warning.
9. Deploy classifier on the Base Stations for the downlink traffic.
1. In the Network Browser, select all base stations; right-click and
select Edit Attributes on any selected node in the scenario space.
Check Apply changes to selected objects.
2. Edit WiMAX Parameters/Classifier Definitions and add one more
row, setting it as in step 3.b.i.
3. Click OK as needed to apply changes to all selected nodes (BS
nodes).
40. OPNET provides a rich set of statistics to examine traffic performance from
the application as well as WiMAX perspective. In this scenario, we collect
global voice application and node level WiMAX connection statistics.
1. Right-click in the scenario space and Choose Individual DES
Statistics.
2. Expand Global Statistics group, and select Voice statistics.
74
3. Expand Node Statistics group, and select WiMAX Connection
statistic.
4. Click OK to accept selections
41. Click on the Configure/Run Discrete Event Simulation button in the
Project Editor. The Configure/Run DES dialog box appears. Notice the
simulation is configured to run for 600 seconds. Click Run to launch the
simulation. When the execution completes, click Close.
42. To view the results Select DES > Results > View Results.
43. To setup VoIP using Session Initiation Protocol (SIP), right click on the
voice server, expand the Edit Attributes (advanced) > SIP > SIP Proxy
Server Parameters and set Proxy Service to Enabled and then Enter the
Server Address as Proxy and Click OK.
FIGURE 6.9. SIP Server Attributes
44. Now select the mobile nodes among which the application is deployed.
Expand Edit Attributes > SIP > SIP UAC Parameters and set SIP UAC
Service to Enabled. Expand Proxy Server Specification and Select the
Proxy Server Name as Proxy. This is the name assigned to the Proxy
Server set in the Voice Server. Apply changes to the selected nodes and
Click OK.
75
FIGURE 6.10. SIP UAC Parameter
6.2.2 Mobile Scenario
With mobility coming into consideration, a new node has to be included to
perform the functionality of the ASN. An ethernet4_slip8_gtwy node is taken from
the object palette and is placed between the backbone and is connected to the
backbone and IP cloud by point to point links. Generic Routing Encapsulation
Tunnels are also to be setup between the WiMAX Base Stations and the ASN
gateway. Generic Routing Encapsulation (GRE) is a tunneling protocol that can
encapsulate wide variety of network layer protocol packet types inside IP tunnels,
creating a virtual point to point links to routers at remote points over an IP
internetwork.
GRE tunnels are designed to be completely stateless. This means that each
tunnel end-point does not keep any information about the state or availability of the
remote tunnel end-point. A consequence of this is that the local tunnel end-point
router does not have the ability to bring the line protocol of the GRE tunnel
interface down if the remote end-point is unreachable. The ability to mark an
interface as down when the remote end of the link is not available is used in order
to remove any routes (specifically static routes) in the routing table that use that
interface as the outbound interface. Specifically, if the line protocol for an interface
is changed to down, then any static routes that point out that interface are removed
from the routing table. This allows for the installation of an alternate (floating)
static route or for policy based routing (PBR) to select an alternate next-hop or
interface.
76
Normally, a GRE tunnel interface comes up as soon as it is configured and
it stays up as long as there is a valid tunnel source address or interface which is up.
The tunnel destination IP address must also be routable. This is true even if the
other side of the tunnel has not been configured. This means that a static route or
PBR forwarding of packets via the GRE tunnel interface remains in effect even
though the GRE tunnel packets do not reach the other end of the tunnel.
The GRE tunnel’s keep alive mechanism is slightly different than for
ethernet or serial interfaces. It gives the ability for one side to originate and receive
keep alive packets to and from a remote router even if the remote router does not
support GRE keep alives. Since GRE is a packet tunneling mechanism for
tunneling IP-in-IP, a GRE IP tunnel packet can be built inside another GRE IP
tunnel packet. For GRE keep alives, the sender pre-builds the keep alive’s response
packet inside the original keep alive request packet so that the remote end only
needs to do standard GRE decapsulation of the outer GRE IP header and then
forward the inner IP GRE packet. This mechanism causes the keep alive’s response
to forward out the physical interface rather than the tunnel interface. This means
that the GRE keep alives response packet is not affected by any output features on
the tunnel interface.
Another attribute of GRE tunnel keep alives is that the keep alive timers on
each side are independent and do not have to match. The problem with the
configuration of keep alives only on one side of the tunnel is that only the router
that has keep alives configured marks its tunnel interface as down if the keep alive
timer expires. The GRE tunnel interface on the other side, where keep alives are
not configured, remains up even if the other side of the tunnel is down. The tunnel
can become a black-hole for packets directed into the tunnel from the side that did
not have keep alives configured. In a large hub-and-spoke GRE tunnel network, it
might be appropriate to only configure GRE keep alives on the spoke side and not
on the hub side. This is because it is often more important for the spoke to discover
that the hub is unreachable and therefore switch to a backup path (Dial Backup for
example).
Before GRE keep alives were implemented, there were only three reasons for a
GRE tunnel to shut down:
o There is no route to the tunnel destination address.
o The interface that anchors the tunnel source is down.
77
o The route to the tunnel destination address is through the tunnel itself.
These three rules (missing route, interface down and mis-routed tunnel
destination) are problems local to the router at the tunnel endpoints and do not
cover problems in the intervening network. For example, these rules do not cover
the case in which the GRE tunneled packets are successfully forwarded, but are
lost before they reach the other end of the tunnel. This causes data packets that go
through the GRE tunnel to be "black holed", even though an alternate route that
uses PBR or a floating static route via another interface is potentially available.
Keep alives on the GRE tunnel interface are used in order to solve this issue in the
same way as keep alives are used on physical interfaces.
Steps to be done to a deployed static WiMAX network to add mobility to it:
1. Change the efficiency setting to use mobility in WiMAX scenario. Go to
View > Show Network Browser. In the network browser pane, right-click
to Edit Attributes on the WiMAX_Config node. Set Efficiency Mode to
Mobility and Ranging Enabled. Click OK.
2. To define the trajectory, go to Topology > Define Trajectory. A window
pops up where the trajectory name is to be entered. Check Coordinates are
relative to the objects and click define path.
3. A new window appears and allows entering the speed of the mobile node.
Click on the starting point of the trajectory and the ending point of the
trajectory the right click and select complete trajectory definition.
4. Select the nodes which will be configured to move. Right Click and select
Edit Attributes (advanced). From the drop down menu of the attribute
trajectory select the trajectory name that was defined.
5. A white line indicating the path of the motion of the mobile node appears.
6. The ASN gateway and the Base Stations are configured to setup the GRE
Tunnel. The configuration of the ASN gateway is as follows:
a. Right click on the ASN gateway -> Edit Attributes (advanced) -> IP
-> IP Routing Parameters -> Interface Information -> IF9 (any other
interface can also be chosen) -> set address field as 192.0.3.1. ->
Click OK to save the settings.
b. Right click on the ASN gateway -> Edit Attributes (advanced) -> IP
-> IP Routing Parameters -> Tunnel interface -> set Number of
78
rows with the number of tunnels to be established. Since we are
trying to develop 7 tunnels, we set the Number of Rows as 7 ->
Click OK to save the settings.
Tunnel Configuration between ASN Gateway and Base Station 1 is as
follows:
1. Right click on the ASN gateway -> Edit Attributes(advanced) > IP -> IP Routing Parameters -> Tunnel interface ->
Tunnel_BS1 -> set Address as 192.1.2.2 -> click OK to save the
settings.
2. Right click on the ASN gateway -> edit Attributes(advanced) > IP -> IP Routing Parameters -> Tunnel interface ->
Tunnel_BS1 -> Tunnel information -> set Tunnel source as
192.0.4.1 and set Tunnel Destination as 192.0.1.1.-> click OK to
save the settings.
3. Right click on the ASN gateway -> edit Attributes(advanced) > IP -> IP Routing Parameters -> Tunnel interface -> Tunnel0 > set Address as 192.1.2.1 and Subnet Mask as Class
C(natural)-> click OK to save the settings.
4. Right click on the ASN gateway -> edit Attributes(advanced) > IP -> IP Routing Parameters -> Tunnel interface -> Tunnel0 > Tunnel Information -> set Tunnel Source as 192.0.1.1 and
Tunnel Destination as 192.0.4.1 -> click OK to save the settings.
The ASN gateway is now configured for the GRE tunnel between
Base Station 1 and the ASN_gateway. Now the Base Station 1 is to
be configured as follows:1. Right
Click
on
the
Base
Station
1
->
Edit
Attributes(advanced) -> IP -> IP Routing Parameters ->
Interface Information -> IF4(any other interfaces may also
be chosen) -> set Address field as 192.0.1.1 and subnet mask
as 255.255.255.0 -> click OK to save the settings.
2. Right Click on the Base Station 1 -> Edit Attributes
(advanced) -> IP -> IP Routing Parameters -> Tunnel
interface -> set Number of rows with the number of tunnels
to be established. Since we are trying to develop 1 tunnels,
79
we set the Number of Rows as 1-> click OK to save the
settings
7. The Base Station 1 is now configured for the GRE Tunnel between Base
Station 1 and the ASN_gateway. To view the GRE Tunnel that has been
developed click View -> Visualize Protocol Configuration -> IP Tunnel
Configuration. Two unidirectional green colored lines will appear between
the ASN gateway and the Base Station1.These are the GRE Tunnels. To
configure the other base stations and the ASN gateway following values are
mapped with the example and set in the same way. The detailed interface
listing can be viewed from Edit -> Edit Objects Using Template -> Click
on the + symbol beside IP and select Tunnel Interface Listing -> Click
Generate. A table giving the details will appear.
TABLE 6.1. Tunnel Interface Listing
Node
Tunnel
Tunnel
Tunnel
name
Address
Subnet
Source IP
Destination
IP
Mask
ASN_GW
Tunnel_BS1 192.1.2.2
Auto
192.0.1.1
192.0.2.1
192.0.1.1
192.0.3.1
192.0.1.1
192.0.4.1
192.0.1.1
192.0.5.1
192.0.1.1
192.0.6.1
192.0.1.1
192.0.7.1
192.0.1.1
192.0.8.1
Assigned
Tunnel_BS2 192.1.3.2
Auto
Assigned
Tunnel_BS3 192.1.4.2
Auto
Assigned
Tunnel_BS4 192.1.5.2
Auto
Assigned
Tunnel_BS5 192.1.6.2
Auto
Assigned
Tunnel_BS6 192.1.7.2
Auto
Assigned
Tunnel_BS7 192.1.8.2
Auto
Assigned
Base
Tunnel0
192.1.2.1
Class C
192.0.2.1
192.0.1.1
Tunnel0
192.1.3.1
Class C
192.0.3.1
192.0.1.1
Station_1
Base
80
Station_2
Base
Tunnel0
192.1.4.1
Class C
192.0.4.1
192.0.1.1
Tunnel0
192.1.5.1
Class C
192.0.5.1
192.0.1.1
Tunnel0
192.1.6.1
Class C
192.0.6.1
192.0.1.1
Tunnel0
192.1.7.1
Class C
192.0.7.1
192.0.1.1
Tunnel0
192.1.8.1
Class C
192.0.8.1
192.0.1.1
Station_3
Base
Station_4
Base
Station_5
Base
Station_6
Base
Station_7
8. After the GRE tunnels are setup, the application demand has to be setup.
9. From the object palette scroll to demands and after expanding demands
select IP_<Voice codec>_Voice. Click on the source node and further click
on the destination node. This creates a unidirectional flow of voice traffic.
Similarly, create the voice traffic from the destination to the source.
10. After the GRE Tunnels are setup and the Application Demands are setup
and mobility is assigned to the nodes the network appears as follows. The
green lines depicts the GRE tunnels that has been setup, the white lines
indicate the path of motion of the mobile nodes and the blue dotted
bidirectional arrows implies the Application Demand traffic configured.
FIGURE 6.11. Deployed Mobile WiMAX Network
11. Select the Demand Statistics from the DES which needs to be plotted and
the scenario is run for 600 sec.
81
6.3 Development of Wireless LAN in OPNET
To deploy a large 7 cell WLAN network and examine the performance of
voice applications in static and mobile conditions the following steps are to be
followed:
Methodology
1. Create a 7 cell WLAN network and connect it to an IP backbone
2. Deploy voice applications between WLAN nodes and configure the voice
server as SIP Server.
3. Examine the performance of voice application in static conditions
4. Deploy trajectories on WiMAX SS nodes to make them mobile.
5. The impact of mobility on application performance is then observed.
6.3.1 Static Scenario:
Steps to develop the static scenario are as follows:
1. Start OPNET Modeler by double-clicking on the OPNET Modeler 14.5
icon.
2. Select File > New.
3. A new window appears. Select Project from the menu and click OK
4. Another window appears and asks for project name and scenario name.
Enter project name as WLAN_OPNET and scenario name as Static.
Uncheck the check box to not use the startup wizard and click OK.
5. Select Topology > Deploy Wireless Network. This launches the Wireless
Network Deployment Wizard.
6. Click Continue and next to move to the Next Tab of technology as shown in
the flowing figure. Select the technology as WLAN Infrastructure and set
the Subscriber node transmission power as 0.005 W, Operational Mode as
802.11b and Data Rate as 11 Mbps. Click next and move to the next Tab of
Topology.
82
FIGURE 6.12. Technology Selection Window
7. In the topology tab enter the cell radius to be 0.1 km and from the drop
down menu select the node placement to be circular. This is because
circular nodes are located equidistantly from the base stations. The settings
are shown in the following figure.
FIGURE 6.13. Topology Selection Window
8. On clicking next the second tab of topology appears as follows. Select the
Subscriber Station as wlan_wkstn_adv from the drop down menu. This
node allows SIP configuration. Click Next.
9. The node mobility tab appears. Since we are deploying a static network
click on the row and click on delete row. Click Next.
10. The configuration summary appears as follows.
83
FIGURE 6.14. Configuration Summary Window
11. Click Finish. The deployed scenario consists of a 7 celled WLAN network
with the 7 access points connected to the backbone.
12. Expand the buffer size by, Right-clicking on any of the mobile nodes, and
then choosing Select Similar Nodes. Right-click on any node again, and
select Edit Attributes on any selected node in the scenario space.
13. Check Apply changes to selected objects.
14. Click on the palette icon (
), scroll up to the ppp_server and drag and
drop it on the work space. Similarly, drag an ip32_cloud and drop it on the
workspace. Rename the ppp_server as the Voice_Server.
15. Now scroll up to the Links_PPP and drag a PPP_SONET_OC1 link to
connect
the
ppp_server
and
the
IP
cloud.
Similarly
select
a
PPP_SONET_OC_12 link and connect the IP cloud to the node called
backbone. The scenario now looks as follows. Right-click in the project
workspace after connecting the link to deactivate the handler.
FIGURE 6.15. Deployed WLAN Network
84
16. Close the Object Palette.
17. From the object palette bring a Application Config and a Profile Config
node to configure the application.
18. Right click on the Application Config and select Edit Attributes
(advanced).
19. Under Application Definition add a row.
20. Enter Application Name as voice.
21. Expand the Description menu select the value of voice menu to edit.
22. Select the encoding scheme.
23. The voice codecs with (silence) implies silence suppression. By default
OPNET takes 65% of the communication period as silence and 35% as the
talk spurt.
24. The Profile Config node is configured as shown in the following figure.
FIGURE 6.16. WLAN Profile Configuration
25. Now with the application and profile configured, we need to deploy the
application.
26. Select Protocols > Applications > Deploy Defined Applications
27. On the right-hand pane, expand Profile:”voice_profile” > Source and
click on Source. On the left-hand pane, while pressing Ctrl, select all
nodes of Cell 2. Once all of these nodes are selected, click on the
assignment arrow (>>) between the two panes.
28. Similarly, on the right-hand pane, expand Application:” voice” > Tier:”
Voice Destination” and click on “Voice Destination.” On the left-hand
85
pane, select all nodes of cell 3, and click on the assignment arrow between
the two panes. At this point, your right-hand pane should look like the
following figure. Click OK.
FIGURE 6.17. WLAN Application Deployment
29. OPNET provides a rich set of statistics to examine traffic performance from
the application as well as WiMAX perspective. In this scenario, we collect
global voice application and node level WiMAX connection statistics.
1. Right-click in the scenario space and Choose Individual DES
Statistics.
2. Expand Global Statistics group, and select Voice statistics.
3. Expand Node Statistics group, and select WiMAX Connection
statistic.
4. Click OK to accept selections
30. Click on the Configure/Run Discrete Event Simulation button in the
Project Editor. The Configure/Run DES dialog box appears. Notice the
simulation is configured to run for 600 seconds. Click Run to launch the
simulation. When the execution completes, click Close.
31. To view the results Select DES > Results > View Results.
32. To setup VoIP using Session Initiation Protocol (SIP), configure the voice
server as done for WiMAX network.
33. Now select the mobile nodes among which the application is deployed and
configure the SIP UAC parameters like those in WIMAX.
86
6.3.2 Mobile Scenario
No new node needs to be added to create a mobile scenario of WLAN. Only
configuring mobility in the mobile nodes and adding Application Demands is all
that is to be done.
1. To define the trajectory, go to Topology > Define Trajectory. A window
pops up where the trajectory name is to be entered. Check Coordinates are
relative to the objects and click define path.
2. A new window appears and allows entering the speed of the mobile node.
Click on the starting point of the trajectory and the ending point of the
trajectory the right click and select complete trajectory definition.
3. Select the nodes which will be configured to move. Right Click and select
Edit Attributes (advanced). From the drop down menu of the attribute
trajectory select the trajectory name that was defined.
4. A white line indicating the path of the motion of the mobile node appears.
5. From the object palette scroll to demands and after expanding demands
select IP_<Voice codec>_Voice. Click on the source node and further click
on the destination node. This creates a unidirectional flow of voice traffic.
Similarly, create the voice traffic from the destination to the source.
6. The WLAN mobile scenario configuration is completed and looks as
follows. The white lines indicate the trajectory and the blue dotted lines
indicate the Application Demands.
FIGURE 6.18. Mobile WLAN Network
7. Select the Demand Statistics from the DES which needs to be plotted and
the scenario is run for 600 sec.
87
6.3 Development of WLAN-WiMAX integrated network in OPNET
6.3.1 Static network
Steps to develop the WLAN-WiMAX integrated network is as follows:1. The integrated static scenario of WLAN-WiMAX integrated network
comprises of two WiMAX Base Stations (wimax_bs_router) connected
via three routers (ethernet4_slip8_gtwy) by PPP_DS3 link. The nodes are
dragged and dropped from the object palette and the links are also taken
from the subhead Links_PPP located in the object palette.
2. A ppp_server is configured as the SIP server as already discussed before
and is connected to the router located in the middle.
3. Two wimax_ss_wlan_router are dragged and dropped one near each base
station. These nodes have the dual stack and behave as a WiMAX
subscriber station and WLAN AP at the same time.
4. IP Auto addressing happens automatically for the WiMAX nodes based on
the association between SS nodes and their servicing BS nodes. In this
case, to associate the wimax_ss_wlan_router to the base station node, a
WiMAX MAC address has to be assigned to the Base Station Node. To do
this right click > Edit Attributes (advanced) > WiMAX Parameters >
BS Parameters > MAC address and then the same MAC address has to
be specified at the wimax_ss_wlan_router node. This can be done as
follows. Right Click > WiMAX Parameters > SS Parameters > BS
MAC Address.
5. The BSS ID of the WLAN also has to be configured manually on the
wimax_ss_wlan_router and the WLAN workstations which are dragged
and dropped into the workspace near the wimax_ss_wlan_router from the
object palette. It is configured on the wimax_ss_wlan_router as follows.
Right click > Edit Attributes (advanced) > Wireless LAN > Wireless
LAN Parameters (IF1 P0) > BSS Identifier. The same has to be put in
the WLAN workstations as follows. Right click > Edit Attributed
(advanced) > Wireless LAN > Wireless LAN Parameters > BSS
Identifier.
6. The Application Config and Profile Config nodes are to be configured as
they were configured in the previous cases.
88
7. The Application then has to be deployed from Protocols > Applications >
Deploy Defined Applications.
8. The above configured scenario appears to be as follows.
FIGURE 6.19. Static WLAN-WiMAX Integrated Network
9. On the WiMAX_Config node, set the Efficiency Mode to Physical Layer
Enabled. Then click OK to accept the changes.
10. Add a classifier definition to the wimax_ss_wlan_router node that directs
voice traffic into the newly configured ertPS connections for uplink traffic.
a. Click on the Value field of WiMAX Parameters/Classifier
Definitions and add one row to the Classifier Definitions table; in
the new row set:
i. Type of SAP to IP
ii. Traffic Characteristics to IP ToS, Equals, Interactive
Voice (6)
iii. Service Class Name to Gold
iv. Click OK once to accept classifier changes.
b. Click OK one final time to apply these changes to the selected
objects.
c. Click Yes to ignore compound attribute warning.
d. Deploy classifier on the Base Stations for the downlink traffic.
i. In the Network Browser, select all base stations; right-click
and select Edit Attributes on any selected node in the
scenario space. Check Apply changes to selected objects.
ii. Edit WiMAX Parameters/Classifier Definitions and add
one more row and configure it as 10a.
iii. Click OK as needed to apply changes to all selected nodes
89
(BS nodes).
11. Configure the SIP Server as the Voice Server is configured in the WiMAX
and WLAN networks and also the SIP UAC parameters are to be
configured like the mobile nodes in the WiMAX and WLAN network.
12. OPNET provides a rich set of statistics to examine traffic performance from
the application as well as WiMAX perspective. In this scenario, we collect
global voice application and node level WiMAX connection statistics.
a. Right-click in the scenario space and Choose Individual DES
Statistics.
b. Expand Global Statistics group, and select Voice statistics.
c. Expand Node Statistics group, and select WiMAX Connection
statistic.
d. Click OK to accept selections
13. Click on the Configure/Run Discrete Event Simulation button in the
Project Editor. The Configure/Run DES dialog box appears. Notice the
simulation is configured to run for 600 seconds. Click Run to launch the
simulation. When the execution completes, click Close.
14. To view the results Select DES > Results > View Results.
6.3.2 Mobile Scenario
1. Deploy a 7-celled WiMAX network and develop a WiMAX network the
way specified in 6.2.1.
2. Drag and drop two WLAN APs (wlan_ethernet_slip4_router) and place it
on either side of the deployed WiMAX network. Connect them to the
Server Backbone using PPP links. These two WLAN APs basically form
the hotspots.
3. 10 wimax_ss_wlan_router behave as handsets here are configured to move
from the coverage of one such hotspot to another moving through the
coverage of WiMAX network.
4. Their trajectory is configured the way already mentioned and Application
demands are configured between these nodes and the WiMAX mobile
nodes under the coverage of Cells 3 and 7.
5. The WiMAX efficiency mode is set to Mobility and Ranging Enabled.
Set up GRE Tunnels between the Base Stations and the ASN Gateway as
90
shown in 6.2.2. This enables the Base Stations to send information about
the mobile node to the ASN Gateway and the ASN Gateway thereby sends
the traffic destined for the mobile node to be sent through the GRE
Tunnels.
6. The Routing information protocol (RIP) is configured throughout the
network and it is by this protocol that the mobile nodes get connectivity
throughout the network.
7. A lower RIP metric is assigned to the WLAN interface while a higher RIP
metric is assigned to the WiMAX interface. This will allow the mobile
node to use the WLAN network whenever both the networks are available.
8. Firstly, both the interfaces are given separate IP. This is done as follows.
Right Click on the mobile node > Edit Attributes (advanced) >IP > IP
Routing Parameters > Interface Information. 2 rows are added here one
for WLAN and the other for WiMAX and configured as follows. The IF0
implies WiMAX Network and the IF1 implies the WLAN network.
TABLE 6.2. Interface Information
Name
Address
Subnet
Routing Protocol MTU (bytes)
Mask
IF0
128.1.1.11
Class C
RIP
1500
IF1
192.0.14.10
Class C
RIP
WLAN
9. To set the RIP metric the following steps are followed. Right click on the
mobile node > Edit Attributes (advanced) > IP Routing Protocols >
RIP Parameters > Interface Information. Set the cost parameter as 3 for
IF0 and 1 for IF1.
10. Set up Application demands as mentioned in 6.2.2 between the mobile
nodes under the access points and the WiMAX Mobile Nodes located under
the coverage of the Base Stations 3 and 7.
11. After the Simulation Setup the scenario appears as follows.
91
FIGURE 6.20. Deployed WLAN-WiMAX Integrated Mobile Network
12. Configure the server as the SIP Server and the mobile nodes as SIP UAC
Clients.
13. Select the Demand Statistics from the DES which needs to be plotted and
the scenario is run for 600 sec.
92
Chapter 7
Comparative Analysis
Chapter Outline:-
 Introduction
 Comparative Results
 Stationary network
 Mobile network
 Conclusion
93
Comparative Analysis
7.1 Introduction
Voice over IP is the expected to be in practice in the next generation
communication networks. The target of this paper is to send voice over IP platform
in the next generation networks and analyse the voice quality. There are various
types of competing networks like WiMAX, Wifi, etc. WiMAX having higher
bandwidth accommodates more users than WLAN but with degraded performance
as shown in this paper. Hence, an integrated network using WiMAX backbone and
WLAN hotspots have been developed with multiple competing traffic using SIP
signalling and analysed to support VoIP. To provide a real life simulation
environment, OPNET 14.5.A modeler is chosen as the simulation tool. The quality
of the service is analysed in all three networks with parameters like average jitter,
average MOS and average packet end to end delay for various voice codecs. The
network performance is also observed with respect to vehicular mobility of 50,
100, and 150km/hr. It is concluded that the WiMAX-WLAN integrated network
accommodates users more than a similar WLAN but less than a similar WiMAX
network. It provides voice quality better than WiMAX and almost like WLAN
which is good and in the mobile scenario the integrated network performs almost
like WiMAX which is much better than WLAN as WLAN cannot support
vehicular mobility.
7.2 Comparative Results
The results obtained in the Chapters 3, 4 and 5 are compared in this section of the
thesis.
7.2.1 Stationary Network
The variation of jitter for the three networks with variation of the voice codecs is
shown in Figure 7.1. Perceived voice quality is considered to be good if the jitter is
zero. As, shown in the figure, the average voice jitter is almost 0 for WiMAX
implying very good quality of voice in comparison to WLAN which has a positive
jitter varying from about 0.0007 to 0.001 seconds. The integrated network shows
jitter variation from about 0.0004 to 0.0006 seconds. For voice codecs G723.1, the
94
average jitter is almost 0 irrespective of the network indicating a good
performance. This is because the bit rate of G723.1 is 6.3 or 5.3 Kbps which results
in generation of small packets. But modem and fax signals cannot be carried by
G723.1 [4]. It can be used only for narrow band communications.
Mean v oice jitter(sec) for v arious codecs
Mean voice jitter (sec)
Integrated Network
0.0012
0.001
0.0008
0.0006
0.0004
0.0002
0
-0.0002
G729
G711
G723.1 G723.1
5.3K
6.3K
W iMAX
G726
G726
16K
24K
Voice Codecs
W LAN
G726
32K
G726
40K
G728
12.8K
G728
16K
Figure. 7.1 Average voice jitter
With silence suppression, the result is different. Like G711, G726 has its roots in
the PSTN network. It is primarily used for international trunks to save bandwidth.
Unlike G711, G726 uses 32Kbps to provide nearly the same quality of voice. This
is because 32 Kbps is the de facto standard.
As shown in Figure 7.2, the average voice jitter is almost 0 for all codecs in both
WLAN and WLAN-WiMAX integrated network where as the WiMAX network
shows a slight deviation for the voice codec G 726. The deviation is highest for
G726 with 32kbps and is about 0.03sec. Though G726 supports data rate of 16 ,
24, and 40Kbps also, the 24 and 16 Kbps channels are used for voice in Digital
Circuit Multiplication Equipment (DCME) and the 40 Kbps is for data modem
signals (especially modems doing 4800 Kbps or higher) in DCME. But G726 with
32 Kbps follows the de facto standard and hence the deviation in jitter is within the
consideration limit.
Mean voice jitter (sec)
using silence suppression
Mean voice jitter
(sec)
Integrated Network
WiMAX
WLAN
3.00E-02
2.00E-02
1.00E-02
0.00E+00
-1.00E-02
G729
G711 G723.1 G723.1 G726 G726 G726
5.3K 6.3K
16K
24K
32K
Voice Codecs
G726 G728
40K 12.8K
G728
16K
Figure. 7.2. Average voice jitter with Silence Suppression
95
Voice Packet end to end
delay(sec)
Voice Packet end to end delay (sec)
5
4
Integrated Network
3
WiMAX
2
WLAN
1
0
G729
G711 G723.1 G723.1
5.3K
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Codecs
Figure. 7.3. Average packet end to end delay
As shown in Figure 7.3, the packet end to end delay for voice without silence
suppression is less than 0.5 seconds for WiMAX where as for WLAN high bit rate
codecs it is very high. This is because the silence periods is also packetised and
send thereby creating huge bandwidth requirement and congestion in the WLAN
network. The integrated network shows a packet end to end delay less than the
WLAN network as the capacity of the network has got enhanced due to the
integration thereby resulting in less congestion. On the other hand, as shown in
figure 7.4, the packet end-to-end delay with silence suppression shows that
WiMAX has a packet end-to-end delay more than WLAN and the integrated
network has the same almost as that of WiMAX network. This is because with
silence suppression the number of packets to be send decreases thereby releasing
the congestion in the WLAN network and also the distance to be traversed by the
packet for WiMAX is much higher than the distance to be covered by the packet in
case of WLAN and even for the integrated network the distance traversed by the
packet is almost is as huge as that of WiMAX. For WLAN the packet end-to-end
delay is about 0.06 seconds except for G 723.1 for which it is about 0.1 seconds
where as for WiMAX and WLAN-WiMAX integrated network, the delay varies
from 0.08 seconds for G 729 and G 711 to 0.13 seconds for G723.1.
Voice Packet end to
end delay(sec)
Voice Packet end to end delay(sec)
using silence suppression
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
Integrated Network
WiMAX
WLAN
G729
G711
G723.1 G723.1
5.3K
6.3K
G726
16K
G726
24K
G726
32K
G726
40K
G728
12.8K
G728
16K
Voice Codecs
Figure. 7.4. Average packet end to end delay with Silence Suppression
96
Figure 7.5 shows the variation of MOS for both WiMAX and WLAN network
with variation of the voice codecs.
Mean Voic e MO S
Value
M e a n V o ic e M O S v a lu e v a ry in g
c o d ec s
Inte g ra te d
W iM A X
WLAN
4
3 .5
3
2 .5
2
1 .5
1
0 .5
0
G 729
G 7 1 1 G 7 2 3 . 1 G 7 2 3 .1 G 7 2 6
5 .3 K
6 .3 K
16K
G 726
24K
G 726
32K
G 726 G 728
40K 12.8K
G 728
16K
Figure. 7.5. Average voice MOS in WiMAX and WLAN and Integrated network
As shown in the figure, the MOS value obtained for WiMAX is above 3, for
WLAN is almost 1 and for the integrated network is about 1.5 except for G 723.1
for which it is about 2.5. This is because G723.1 is a low bit rate codec. In case of
WiMAX it has packet end to end delay higher than the other codecs hence lower
MOS. On the other hand it has low bandwidth requirement and hence less packets
are dropped for this codec in WLAN network. Hence higher MOS compared to the
other codecs.
Mean Voice MOS value varying
codecs with s ilence s uppres s ion
Mean Voic e MO S Value
Integrated
WiMAX
WLAN
4
3.5
3
2.5
2
1.5
1
0.5
0
G729 G711 G723.1 G723.1 G726 G726 G726 G726 G728 G728
5.3K 6.3K 16K
24K
32K
40K 12.8K 16K
Voice C odecs
Figure. 7.6. Average voice MOS with Silence Suppression
The result with silence suppression is totally different. As shown in Figure 7.6,
WLAN shows better performance among the three networks with a MOS value of
above 3.5 except for G711 for which it is about 3 and for G723.1 for which it is
about 2.5. This is because with the silence suppression the number of packets in
97
the network decreases thereby releasing the congestion in the WLAN network.
This decreases the amount of packet dropped and thereby increases the MOS
value. The codec G723.1 is a low bit rate codec having bit rate of 5.3 kbps and
6.3kbps and the packet end to end delay is larger for G 723.1 in comparison to the
other codecs. Hence the MOS decreases.
7.2.2 Mobile Network
Figure 7.7 shows the variation of jitter for all three networks in the mobile
scenario. The mobile nodes are configured to move at a speed of 50, 100,150
km/hr. As shown in the figure, the average jitter for WLAN network is very high
compared to the WiMAX network. The WLAN-WiMAX integrated network
shows variation of jitter almost like the WiMAX which is very low compared to
WLAN implying very good performance. The average performance of the
networks with respect to speed is almost same. Among the integrated network and
the WiMAX network, the WiMAX network performs slightly better than the
integrated network. The same with silence suppression could not be noted due to
the limitation of OPNET.
Average Jitter
G723.1
G726
G728
G729
WiMAX
WLAN
150km/hr
100km/hr
50km/hr
150km/hr
100km/hr
50km/hr
150km/hr
100km/hr
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
50km/hr
Sec
G711
Intergrated
Network
Figure 7.7 Average Jitter in mobile network
As shown in figure 7.8 the packet end-to-end delay for WiMAX and
WLAN-WiMAX is negligible compared to WLAN network. The delay exhibited
by the WLAN is of the order of minutes which make the network practically
unusable. Figure 7.9 shows the variation of delay between the WiMAX network
and the integrated network which is almost zero with respect to WLAN. Among
98
the two networks again WLAN is found to perform best and the integrated network
performs in a slightly degraded way. Also it is observed that delay increases with
the speed and G 711 has highest delay among other codecs for integrated network.
This is because G 711 is the highest bit rate codec and has a data rate of 64Kbps.
As the mobile nodes move, they get handed over both horizontally and vertically.
With high data rate, packets get dropped more during the handover. Any packet
dropped by the lower layer is perceived as delay by the upper layers. Hence, the
delay of G 711 is exhibited in figure 7.9.
Packet End-to-End Delay
G711
G723.1
G726
G728
G729
180
160
140
Sec
120
100
80
60
40
20
WiMAX
WLAN
150km/hr
100km/hr
50km/hr
150km/hr
100km/hr
50km/hr
150km/hr
100km/hr
50km/hr
0
Integrated
Network
Figure 7.8 Average Packet End-to-End Delay for three networks
Packet End-to-End Delay
G711
G723.1
G726
G728
G729
0.3
0.25
sec
0.2
0.15
0.1
0.05
0
50km/hr
100km/hr 150km/hr
50km/hr
WiMAX
100km/hr 150km/hr
Integrated
Network
Figure 7.9 Average Packet End-to-End Delay for two networks
99
7.3 Conclusion
In this chapter the simulated results of a VoIP application over a WiMAX
based BWA network, a WLAN 802.11b network and a WLAN-WiMAX integrated
network in OPNET 14.5.A are presented. Multiple competing traffic sources using
SIP signalling over the networks are simulated and the traffic traces and
measurements obtained from the simulation are presented.
Overall it can be concluded that to support increasing number of users,
silence suppression has to be used. With silence suppression introduced in the
network, WLAN is observed to perform best in stationary condition but WLAN
cannot support vehicular mobility. To support vehicular mobility WiMAX has to
be introduced. It is observed from the mobile scenario that WiMAX performs
much better than WLAN. The integrated network in the static scenario performs
better than WiMAX and almost like WLAN and in the mobile scenario; the
integrated network performs almost like WiMAX which is much better than
WLAN. Thus, we see that the integrated network gives the best performance in
both static and mobile scenario.
The mobile scenario with silence suppression could not be simulated due to
the limitations of OPNET. [45] states that the capacity of mobile WiMAX is
doubled when silence suppression is used. Hence, as presented in this thesis, if an
integrated network is developed comprising of WiMAX backbone and WLAN
hotspot, it is expected to be competent enough to provide optimal voice quality
with optimal network capacity.
100
Chapter 8
Conclusion and Future Work
Chapter Outline: Conclusion
 Future Work
101
Conclusion and Future Work
8.1 Conclusion
The variation in performance of a WiMAX network deployed over
suburban areas of terrain types varying from hilly terrain with moderate to heavy
tree density to flat terrain with light tree density for VoIP application with mobile
nodes moving at a speed of 50 km/hr is investigated. The investigation evaluated
the effect of varying suburban terrain; on the throughput and packet end to end
delay performance of Mobile WiMAX. The reduction in Line-Of-Sight due to the
terrain impacts the throughput directly by increasing the attenuation and diffraction
losses, and indirectly affects the packet end to end delay by causing nodes to
initiate network registration more frequently under fluctuating cell coverage. Thus,
it is concluded that any experiment related to wireless network should take into
consideration the terrain on which the network is deployed because this affects the
network performance substantially.
In this project extensive simulation based performance analysis for VoIP
application is done over a WiMAX based BWA network, a WLAN (IEEE
802.11b), and also on a WLAN-WiMAX integrated network. For this OPNET
14.5.A and OPNET 15.0.A simulation platform have been used. Multiple
competing traffic sources using SIP signalling over the networks are generated and
the trace of traffic and measurements for different performance parameters
discussed in this paper are obtained.
Close observation of the results of static networks reveal that
WiMAX network performs better on the basis of jitter, MOS and packet end-toend delay than conventional WLAN 802.11b in case of voice applications when no
silence suppression is considered, i.e. with no bar on the bandwidth usage, but
when more users are to be accommodated, bandwidth becomes a constraint and
hence silence suppression has to be used. With silence suppression, WLAN
provides a better voice quality than WiMAX. Though, the WiMAX network
provides high capacity, degradation of the voice quality is observed i.e. reflected
from the MOS value. WLAN on the other hand has less capacity but provides a
better voice quality. Hence, the integrated deployment of WiMAX and WLAN is
102
expected to be competent enough to provide optimal voice quality with optimal
network capacity.
To support increasing number of users, silence suppression has to be used.
With silence suppression introduced in the network, WLAN is observed to perform
best in stationary condition but WLAN cannot support vehicular mobility. To
support vehicular mobility WiMAX has to be introduced. It is observed from the
mobile scenario that WiMAX performs much better than WLAN. The integrated
network in the static scenario performs better than WiMAX and almost like
WLAN and in the mobile scenario; the integrated network performs almost like
WiMAX which is much better than WLAN. Thus, we see that the integrated
network gives the best performance in both static and mobile scenario.
The mobile scenario with silence suppression could not be simulated due to
the limitations of OPNET. [45] states that the capacity of mobile WiMAX is
doubled when silence suppression is used. Hence, as presented in this thesis, if an
integrated network is developed comprising of WiMAX backbone and WLAN
hotspot, it is expected to be competent enough to provide optimal voice quality
with optimal network capacity.
8.2 Future Work
This thesis has worked on the performance of WLAN 802.11b and
WiMAX 802.16 networks for voice application. The work done in this thesis can
further be extended to observe quite a number of variations. Video traffic when
configured in the WLAN 802.11b network is observed to support hardly 2 to 3
users per access points. Hence, increasing network capacity can be a future step for
video applications. The performance of the networks for varying users can also be
a work in the future scope. Also the network performance for other applications
like file transfer and web browsing can be an extension of the work.
The integrated network used in this work has been configured using
WiMAX (IEEE 802.16) and WLAN (IEEE 802.11b). This part of the work can be
further extended by adding 3G network to the existing infrastructure. OPNET
15.0.A presently does not support the coexistence of WiMAX or WLAN and 3G. It
may be supported in Qualnet or in future releases of OPNET and thereby its
performance can be observed.
103
Voice Activity Detection is a technique required for silence suppression in
VoIP and is not currently supported in the application demands of OPNET. Also
the plot for Mean Optimal Score (MOS) calculation is not presently supported by
OPNET for WLAN-WiMAX integrated mobile scenario. This is a limitation of
OPNET 15.0.A. which may be overcome by modifying some codes of OPNET
modeler and can be in the scope of future work.
For representing the WLAN technology IEEE 802.11b has been considered
as it is low cost and widely deployed and accepted by the public, but IEEE 802.11b
supports highest data rate of 11 Mbps. On the other hand IEEE 802.11a supports
high data rate of 54 Mbps but is still in the development stage and not widely used.
Hence, performance of the WiMAX, WLAN and the WiMAX-WLAN integrated
network can be observed for IEEE 802.11a or the most recently developed IEEE
802.11n with respect to this work as a future work.
104
Bibliography:
[1]
B. P. Crow, I. Widjaja, J. G. Kim, “IEEE 802.11 Wireless Local Area
Networks”, Proceedings of IEEE Communications, 1997
[2]
D. Malone, P. Clifford, D.J. Leith, “On Buffer Sizing for Voice in 802.11
WLANs”, Proceedings of IEEE Communications Letters, 2007
[3]
http://voip.biz-news.com/news/tags/en_US/VoIP
[4]
“Smart Mobile™: Next-Generation Wlan Architecture For High Performance
Networks”, A Trapeze Networks White Paper, November 2006.
[5]
D.Y. Chen, S. Garg, M. Kappes, and K. S. Trivedi, “Supporting VBR VoIP
traffic with IEEE 802.11 WLAN in PCF mode”, Proceedings of
OPNETWork 2002, Washington D. C., August 2002.
[6]
F.E.Retnasothie, M.K. Ozdemir , T.Y’ Cek , H. Celebi , J. Zhang, and R.
Muththaiah, “Wireless IPTV over WiMAX: Challenges and Applications”,
Proceedings of IEEE WAMICON, 2006
[7]
L. X. Cai, X. (Sherman) Shen, J. W. Mark, L. Cai, Y. Xiao, “Voice Capacity
Analysis of WLAN With Unbalanced Traffic”, Proceedings of IEEE
Transactions on Vehicular Technologies, Vol. 55, No. 3, May 2006
[8]
N.Nagarajan , Rajeev, B. Kaarthick, “Performnance analysis of Video
Conferencing and Multimedia application Services over WiMAX”,
Proceedings of IEEE International Advance Computing Conference (IACC
2009), Patiala, India, 6-7 March 2009
[9]
G. Chu, D.Wang, S. Mei, “A QoS Architecture for the MAC protocol of
IEEE 802.16 BWA System,” Proceedings of IEEE 2002 ICC, Circuits and
Systems, vol. 1, July 2002, pp. 435 -439.
[10]
B.Angelov, B.Rao, “The Progression of WiMAX Toward a Peer-to- Peer
Paradigm Shift”, Annual Review of Communications, Vol 59, 2007
[11]
http://whirlpool.net.au/wiki/?tag=VoIP_Codecs
[12]
http://en.wikipedia.org/wiki/G.726
[13]
A. Trad, F. Munir, H. Afifi, “Capacity Evaluation of VoIP in IEEE 802.11e
WLAN Environment”, IEEE CCNC 2006 Proceedings, 2006
[14]
http://wiki.snom.com/Bandwidth_requirements_and_QoS
[15]
O.Hersent, J.Petit, D.Gurle, IP Telephony- Deploying Voice-Over-IP
Protocols, John Wiley & Sons, Ltd,2005
[16]
www.cs.wustl.edu/~jain/cse574-06/ftp/wimax_voip/index.html
[17]
ITU, G.114 Delay Recommendations (1996).
[18]
B. Preveze, H. T. Eyyuboğlu, “A Simulated Analysis of Delay Factors for
IP”, Int. Conf. on Intelligent Knowledge Systems (IKS-2004), August 2004
[19]
P.C. Metha, and S. Udani, (2001) “Overview of Voice over IP”, Technical
Report
MS-CIS-01-31.
From
http://citeseer.ist.psu.edu/mehta01overview.html.
[20]
ETSI TR 101 329-7 2.1.1 (2002-02), Telecommunications and Internet
Protocol Harmonization over networks release 3: End to end quality of
service in TIPHON systems Technical Report.
[21]
Cisco Systems White Paper (2003). Understanding Delay in Packet Voice
Networks.
From
http://www.cisco.com/warp/public/788/voip/delaydetails.pdf.
[22]
Brans, T., De Keyser, T., Peirs, C. and Pollin, S. (2001), “Voice Over IP”,
December 2001.
[23]
Cisco Systems White Paper (2002), Understanding Codecs: Complexity,
Hardware Support, MOS, and Negogiation.
[24]
Wright, D.J. (2001). Voice Over Packet Networks: John Wiley & Sons, Ltd.
[25]
Iti Saha Misra, “Wireless Communications and Networks 3G and beyond”,
Tata McGraw Hill Education Private Limited, 2009, New Delhi.
[26]
V.Erceg et al., “An Empirically Based Path Loss Model for Wireless
Channels in Suburban Environments”, IEEE JSAC, vol. 17, July 1999, pp.
1205-1222
[27]
Jeich Mar, Chin-Chung Ko, Chung-Haw Li and Shao-En Chen, “CELL
PLANNING AND CHANNEL THROUGHPUT OF MOBILE WiMAX AT
2.5 GHz”, vol. 32, No. 5, Journal of the Chinese Institute of Engineers, 2009,
pp.585–597.
[28]
ITU-R, M.1225 “GUIDELINES FOR EVALUATION OF RADIO
TRANSMISION TECHNOLOGIES FOR IMT-2000”
[29]
Aphiraksatyakul,D., Boon-Chong Seet, Chiew-Tong Lau, “Evaluation of
Terrain Effects on Mobile WiMax in a Vehicular Environment”,ITS
Telecommunications 2008, pp.379-383, October 2008.
[30]
Deepak Pareek, WiMAX Taking wireless to the MAX, Auerbach
Publications, Taylor and Francis Group, Boca Raton, New York, 2006
[31]
Emir Halepovic, Majid Ghaderi, Carey Williamson, “Multimedia Application
Performace on a WiMAX Network”, SPIE Multimedia Computing and
Networking 2009, Vol. 7253, San Jose, CA, January 2009.
[32]
IEEE Standard 802.16 – 2004. Part 16: Air Interface for fixed broadband
wireless access systems. October 2004.
[33]
IEEE Standard 802.16e – 2005. Part 16: Air Interface for fixed and mobile
broadband wireless access systems – Amendment for physical and Medium
Access Control layers for combined fixed and mobile operation in licensed
band. December 2005.
[34]
http://www.voip-info.org/wiki/view/ITU+G.726S.
[35]
http://en.wikipedia.org/wiki/G723.1
[36]
IEEE Press, “Mobile WiMAX”, John Wiley & Sons, Ltd
[37]
Bong-Ho Kim, Jungnam Yun, and Yerang Hur, Posdata America, Chakchai
So-In, Raj Jain and Abdel-Karim Al Tamimi, Washington University, St.
Louis, “Capacity Estimation and TCP Performance Enhancement over
Mobile WiMAX”, IEEE Communications Magazine, June 2009, pp 132-141.
[38]
Jari Korhonen, Ye Wang, School of Computing, National University of
Singapore, “Effect of Packet Size on Loss Rate and Delay in Wireless
Links”, IEEE Communications Society, WCNC, 2005, pp 1608-1613.
[39]
www.wimax.com
[40]
IEEE Std 802.11-1999, Part 11: Wireless LAN MAC and physical layer
specifications. Reference number ISO/IEC 8802-11:1999(E).
[41]
M. Veeraraghavan, N. Cocker, and T. Moors. Support of voice services in
IEEE 802.11 Wireless LANs. In Proceedings of IEEE INFOCOM 2001,
Alaska, April 2001.
[42]
“WiMAX and WiFi Together: Deployment models and User Scenarios”,
Intel and Motorola.
[43]
M.F. Finneran, “A comparison of Technologies, Markets and Bussiness
Plans”, dBrn Associates, Inc, June 2004.
[44]
The OPNET website. http://www.opnet.com.
[45]
Chakchai So-In, Raj Jain, Abdel-Karim Tamimi, “Capacity Evaluation of
IEEE 802.16e Mobile WiMAX”, Journal of Computer Systems Networks,
Networks and Communications, Volume 2010.