wireless mesh networking architectures, protocols and standards

He alsoserves as co-editor for several books: Resource, Mobility and SecurityManagement in Wireless Networks and Mobile Communications;Wireless Mesh Networking: Architectures, Protocols

MESH NETWORKING

Architecting the Telecommunication

Evolution: Toward Converged Network

Context-Aware Pervasive Systems:

Architectures for a New Breed of

Introduction to Mobile Communications:

Technology, Services, Markets

Performance Modeling and Analysis of

Bluetooth Networks: Polling, Scheduling,

and Traffic Control

Jelena Misic and Vojislav B Misic

Resource, Mobility, and Security

Management in Wireless Networks

and Mobile Communications

Yan Zhang, Honglin Hu, and Masayuki Fujise

ISBN: 0-8493-8036-7

Security in Distributed, Grid, Mobile, and Pervasive Computing

Yang Xiao ISBN: 0-8493-7921-0

TCP Performance over UMTS-HSDPA Systems

Mohamad Assaad and Djamal Zeghlache ISBN: 0-8493-6838-3

Testing Integrated QoS of VoIP:

Packets to Perceptual Voice Quality

Vlatko Lipovac ISBN: 0-8493-3521-3

The Handbook of Mobile Middleware

Paolo Bellavista and Antonio Corradi ISBN: 0-8493-3833-6

Traffic Management in IP-Based Communications

Trinh Anh Tuan ISBN: 0-8493-9577-1

Understanding Broadband over Power Line

Gilbert Held ISBN: 0-8493-9846-0

Understanding IPTV

Gilbert Held ISBN: 0-8493-7415-4

WiMAX: A Wireless Technology Revolution

G.S.V Radha Krishna Rao, G Radhamani ISBN: 0-8493-7059-0

WiMAX: Taking Wireless to the MAX

Deepak Pareek ISBN: 0-8493-7186-4

Wireless Mesh Networking: Architectures, Protocols and Standards

Yan Zhang, Jijun Luo and Honglin Hu ISBN: 0-8493-7399-9

Wireless Mesh Networks

Gilbert Held ISBN: 0-8493-2960-4

AUERBACH PUBLICATIONS

www.auerbach-publications.com

To Order Call: 1-800-272-7737 • Fax: 1-800-374-3401

E-mail: orders@crcpress.com

Boca Raton New York Auerbach Publications is an imprint of the Taylor & Francis Group, an informa business

WIRELESS

MESH NETWORKING

Architectures, Protocols and Standards

Edited by Yan Zhang • Jijun Luo • Honglin Hu

Auerbach Publications

Taylor & Francis Group

6000 Broken Sound Parkway NW, Suite 300

Boca Raton, FL 33487-2742

© 2007 by Taylor & Francis Group, LLC

Auerbach is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number-10: 0-8493-7399-9 (Hardcover)

International Standard Book Number-13: 978-0-8493-7399-2 (Hardcover)

This book contains information obtained from authentic and highly regarded sources Reprinted material is quoted with permission, and sources are indicated A wide variety of references are listed Reasonable efforts have been made to publish reliable data and information, but the author and the publisher cannot assume responsibility for the validity of all materials or for the conse- quences of their use

No part of this book may be reprinted, reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming, and recording, or in any information storage or retrieval system, without written permission from the publishers.

For permission to photocopy or use material electronically from this work, please access www copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc (CCC)

222 Rosewood Drive, Danvers, MA 01923, 978-750-8400 CCC is a not-for-profit organization that provides licenses and registration for a variety of users For organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and

are used only for identification and explanation without intent to infringe.

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com

and the Auerbach Web site at

http://www.auerbach-publications.com

Editors viiContributors ixPreface xiiiPART I: ARCHITECTURES

1 Wireless Mesh Networks: Issues and Solutions 3

B.S MANOJ AND RAMESH R RAO

2 Multiradio Multichannel Mesh Networks 49

RAJIV VIJAYAKUMAR, ARINDAM DAS, SUMIT ROY, AND HUI MA

3 IEEE 802.11-Based Wireless Mesh Networks 79

ASHISH RANIWALA, RUPA KRISHNAN, AND TZI-CKER CHIUEH

PART II: PROTOCOLS

4 Routing in Wireless Mesh Networks 113

MICHAEL BAHR, JIANPING WANG, AND XIAOHUA JIA

MICHELLE X GONG, SHIWEN MAO, SCOTT F MIDKIFF, AND

BRIAN HART

6 Security in Wireless Mesh Networks 183

RAINER FALK, CHIN-TSER HUANG, FLORIAN KOHLMAYER,

AND AI-FEN SUI

7 Scalability in Wireless Mesh Networks 225

JANE-HWA HUANG, LI-CHUN WANG, AND CHUNG-JU CHANG

8 Load Balancing in Wireless Mesh Networks 263

B.S MANOJ AND RAMESH R RAO

9 Cross-Layer Optimization for Scheduling

in Wireless Mesh Networks 297

VASILIS FRIDERIKOS, KATERINA PAPADAKI,

DAVID WISELY, AND HAMID AGHVAMI

Wireless Mesh Networks 329

ENRICO MASALA, ANTONIO SERVETTI, AND

JUAN CARLOS DE MARTIN

Wireless Mesh Networks 361

HONGLIN HU, JIJUN LUO, AND XIAODONG ZHANG

PART III: STANDARDIZATION AND ENABLING

TECHNOLOGIES

Networks in IEEE 802.11s 391

SHAH I RAHMAN

13 IEEE 802.16 WiMAX Mesh Networking 425

YAN ZHANG, KEAN-SOON TAN, PENG-YONG KONG,

JUN ZHENG, AND MASAYUKI FUJISE

Spectrum Management 467

CLEMENS KLOECK, VOLKER BLASCHKE, HOLGER JAEKEL,

FRIEDRICH K JONDRAL, DAVID GRANDBLAISE,

JEAN-CHRISTOPHE DUNAT, AND SOPHIE GAULT

Emergency Management and Market Analysis 509

CARLES GOMEZ, PAU PLANS, MARISA CATALAN, JOSEP LLUIS FERRER, JOSEP PARADELLS, ANNA CALVERAS, JAVIER RUBIO, AND DANIEL ALMODOVAR

Safety and Disaster Recovery Applications 545

MARIUS PORTMANN

Index 577

Yan Zhang received a Ph.D degree from the School of Electrical &Electronics Engineering, Nanyang Technological University, Singapore.From August 2004 to May 2006, he worked with NICT Singapore,National Institute of Information and Communications Technology(NICT) From August 2006, he has worked with Simula ResearchLaboratory, Norway (http://www.simula.no/ )

Dr Zhang is on the editorial board of International Journal

of Network Security and is currently serving as the series editor forthe book series ‘‘Wireless Networks and Mobile Communications’’(Auerbach Publications, CRC Press, Taylor & Francis Group) He alsoserves as co-editor for several books: Resource, Mobility and SecurityManagement in Wireless Networks and Mobile Communications;Wireless Mesh Networking: Architectures, Protocols and Standards;Millimeter-Wave Technology in Wireless PAN, LAN and MAN; Dis-tributed Antenna Systems: Open Architecture for Future WirelessCommunications; Security in Wireless Mesh Networks; Wireless Metro-politan Area Networks: WiMAX and Beyond; Wireless Quality-of-Service: Techniques, Standards; and Applications; BroadbandMobile Multimedia: Techniques and Applications; Internet of Things:From RFID to the Next-Generation Pervasive Networked Systems; andHandbook of Research on Wireless Security

He serves as program co-chair for IEEE PCAC’07, special trackco-chair for ‘‘Mobility and Resource Management in Wireless/MobileNetworks’’ in ITNG 2007, special session co-organizer for ‘‘WirelessMesh Networks’’ in PDCS 2006, is a member of the Technical ProgramCommittee for IEEE AINA 2007, IEEE CCNC 2007, WASA’06, IEEEGLOBECOM’2006, IEEE WoNGeN’06, IEEE IWCMC 2006, IEEE IWCMC

2005, ITST 2006, and ITST 2005 His research interests includeresource, mobility, energy and security management in wireless networks

and mobile computing He is a member of IEEE and IEEE ComSoc.Email: yanzhang@ieee.org

Jijun Luo received his master of engineering (M Eng.) degree fromShandong University, China in 1999 and a master of science (M Sc.)degree from Munich University of Technology, Germany in 2000 Hejoined Siemens in 2000 and pursued his doktor-ingenieur (Dr Ing.)degree at RWTH Aachen University, Germany Until now he haspublished more than 100 technical papers, co-authored three booksand holds many patents His technical contributions mainly coverhis findings on wireless communication system design, radio protocol,system architecture, radio resource management, signal processing,coding and modulation technologies He is active in internationalacademy and research activities He serves as invited reviewer ofIEEE Transactions on Vehicular Technology, IEEE Communica-tions Magazine, IEEE Wireless Communications Magazine, EURASIPJournals, Frequenz, etc He has been nominated as session chairand reviewer of many high-level technical conferences organized byIEEE and European research organizations He is leading Europeanresearch projects and is active in international industrial standardiza-tion bodies His main interests are transmission technologies, radioresource management, reconfigurability (software-defined radio) andradio system design He is a member of IEEE Email: jesse.luo@ieee.orgHonglin Hu received his Ph.D degree in communications and infor-mation systems in January 2004 from the University of Science andTechnology of China (USTC), Hefei, China From July 2004 to January

2006, he was with Future Radio, Siemens AG Communications inMunich, Germany Since January 2006, he has been with the ShanghaiResearch Center for Wireless Communications (SHRCWC), which isalso known as the International Center for Wireless CollaborativeResearch (Wireless CoRe) Meanwhile, he serves as the an associateprofessor at the Shanghai Institute of Microsystem and InformationTechnology (SIMIT), Chinese Academy of Science (CAS) Dr Hu ismainly working for international standardization and other collabora-tive activities He is a member of IEEE, IEEE ComSoc, and IEEE TCPC

In addition, he serves as a member of Technical Program Committeefor IEEE WirelessCom 2005, IEEE ICC 2006, IEEE IWCMC 2006, IEEEICC 2007, IEEE/ACM Q2SWinet 2006 Since June 2006, he has served,

on the editorial board of Wireless Communications and MobileComputing, John Wiley & Sons Email: hlhu@ieee.org

Wireless Networks Group

Technical University of Catalonia

Barcelona, Spain

Marisa Catalan

Wireless Networks Group

Technical University of Catalonia

Barcelona, Spain

Chung-Ju ChangDepartment of CommunicationEngineering

National Chiao-Tung UniversityHsinchu, Taiwan

People’s Republic of China

Tzi-Cker ChiuehDepartment of ComputerScience

Stony Brook UniversityStony Brook, New York

Arindam DasDepartment of ElectricalEngineering

University of Washington,Seattle, Washington

Juan Carlos De MartinComputer and Control EngineeringDepartment

Politecnico di TorinoTorino, Italy

Jean-Christophe DunatMotorola LaboratoriesGif sur Yvette, France

Rainer Falk

Siemens AG

Corporate Technology

Munich, Germany

Josep Lluis Ferrer

Wireless Networks Group

Technical University of Catalonia

National Institute of Information

and Communications Technology

Wireless Networks Group

Technical University of Catalonia

Chin-Tser HuangDepartment of Computer Scienceand Engineering

University of South CarolinaColumbia, South Carolina

Jane-Hwa HuangDepartment of CommunicationEngineering

National Chiao-Tung UniversityHsinchu, Taiwan

People’s Republic of China

Holger JaekelNachrichtentechnikUniversita¨t KarlsruheKarlsruhe, Germany

Xiaohua JiaDepartment of Computer ScienceCity University of Hong KongKowloon, Hong Kong

Friedrich K JondralNachrichtentechnikUniversita¨t KarlsruheKarlsruhe, Germany

Clemens KloeckNachrichtentechnikUniversita¨t KarlsruheKarlsruhe, Germany

Florian KohlmayerSiemens AG

Corporate TechnologyMunich, Germany

Department of Computer Science

Stony Brook University

Stony Brook, New York

Josep ParadellsWireless Networks GroupTechnical University of CataloniaBarcelona, Spain

Pau PlansWireless Networks GroupTechnical University

of CataloniaBarcelona, SpainMarius PortmannSchool of ITEEUniversity of QueenslandBrisbane, AustraliaShah I RahmanCisco SystemsSan Jose, California

Ashish RaniwalaDepartment of ComputerScience

Stony Brook UniversityStony Brook, New York

Ramesh R RaoDepartment of Electrical andComputer EngineeringUniversity of CaliforniaSan Diego, CaliforniaSumit Roy

Department of ElectricalEngineering

University of WashingtonSeattle, Washington

Department Information Security

Beijing, People’s Republic of China

National Chiao-Tung UniversityHsinchu, Taiwan

People’s Republic of ChinaDavid Wisely

Mobility Research, BT GroupIpswich, United KingdomXiaodong ZhangShanghai Research Center forWireless CommunicationsShanghai, People’s Republic

of ChinaYan ZhangSimula Research LaboratoryOslo, Norway

Jun ZhengComputer Science DepartmentQueens College-City UniversityFlushing, New York

Wireless Mesh Networks (WMN) are believed to be a highly promisingtechnology and will play an increasingly important role in futuregeneration wireless mobile networks WMN is characterized bydynamic self-organization, self-configuration and self-healing toenable quick deployment, easy maintenance, low cost, high scalabilityand reliable services, as well as enhancing network capacity, connect-ivity and resilience Due to these advantages, international standard-ization organizations are actively calling for specifications for meshnetworking modes, e.g., IEEE 802.11, IEEE 802.15, IEEE 802.16 andIEEE 802.20 As a great extension to the ad hoc network, WMN isbecoming an important mode complementary to the infrastructure-based wireless networks The experiences obtained from studyingand deploying WMN provide us knowledge and reference to thefuture networks evolution

Wireless Mesh Networking: Architectures, Protocols and Standardsprovides a comprehensive technical guide covering introductoryconcepts, fundamental techniques, recent advances and open issues

in wireless mesh networks It focuses on concepts, effective protocols,system integration, performance analysis techniques, simulation,experiments, and future directions It explores various key challenges,diverse scenarios and emerging standards such as those for capacity,coverage, scalability, extensibility, reliability, and cognition This vol-ume contains illustrative figures and complete cross-referencing onrouting, security, spectrum management, medium access, cross-layeroptimization, load-balancing, multimedia communication, MIMO, andsmart antennas, etc It also details information on particular techniquesfor efficiently improving the performance of a wireless mesh network

This book is organized in three parts:

& Part I: Architectures

& Part II: Protocols

& Part III: Standardization and Enabling Technologies

In Part I, WMN fundamentals are briefly introduced as are varioustypes of network architecture Part II concentrates on the techniquesnecessary to enable a complete, secure and reliable wireless network,including routing, security, medium access control (MAC), scalability,load balancing, cross layer optimization, scheduling, multimedia com-munication, MIMO (or multiple antenna system) Part III exploresstandardization activities and particular mesh network specifications

in the emerging standards, for instance, mesh mode in the IEEE 802.11Wireless LAN and in the IEEE 802.16 WiMAX In addition, the appli-cations of mesh networks in emergency management, and publicsafety are exploited

This book has the following salient features:

& Provides a comprehensive reference on state-of-the-arttechnologies for wireless mesh networks

& Identifies basic concepts, techniques, advanced research topicsand future directions

& Contains illustrative figures that enable easy understanding ofwireless mesh networks

& Allows complete cross-referencing via the broad coverage ofdifferent layers of protocol stacks

& Details particular techniques for efficiently improving the formance of wireless mesh networks

per-This book can serve as a useful reference for students, educators,faculties, telecom service providers, research strategists, scientists,researchers, and engineers in the fields of wireless networks andmobile communications

We would like to acknowledge the effort and time invested by allcontributors for their excellent work They were extremely profes-sional and cooperative Special thanks go to Richard O’Hanley, KimHackett, Catherine Giacari, Glenon Butler and Jessica Vakili of Taylor

& Francis Group for their support, patience and professionalismfrom the beginning to the final stage We are grateful for SuryakalaArulprakasam for her great efforts during the typesetting period Last

but not least, a special thank you to our families and friends for theirconstant encouragement, patience and understanding throughout thisproject.

Yan Zhang, Jijun Luo,and Honglin Hu

PART I

ARCHITECTURES

WIRELESS MESH NETWORKS: ISSUES AND

SOLUTIONS

B.S Manoj and Ramesh R Rao

CONTENTS

1.1 Introduction 5

1.2 Comparison between wireless ad hoc and mesh networks 6

1.3 Challenges in wireless mesh networks 8

1.3.1 Throughput capacity 9

1.3.2 Throughput fairness 10

1.3.3 Reliability and robustness 12

1.3.4 Resource management 13

1.4 Design issues in wireless mesh networks 13

1.4.1 Network architectural design issues 14

1.4.1.1 Flat wireless mesh network 14

1.4.1.2 Hierarchical wireless mesh network 14

1.4.1.3 Hybrid wireless mesh network 14

1.4.2 Network protocol design issues 15

1.4.2.1 Physical layer design issues 15

1.4.2.2 Medium access control layer 16

1.4.2.3 Network layer 16

1.4.2.4 Transport layer 17

1.4.2.5 Application layer 17

1.4.2.6 System-level design issues 17

4

3

1.5 Design issues in multiradio wireless mesh networks 18

1.5.1 Architectural design issues 18

1.5.2 Medium access control design issues 19

1.5.3 Routing protocol design issues 20

1.5.4 Routing metric design issues 21

1.5.5 Topology control design issues 22

1.6 Link layer solutions for multiradio wireless mesh networks 23

1.6.1 Multiradio unification protocol 24

1.7 Medium access control protocols for multiradio wireless mesh networks 28

1.7.1 Multichannel CSMA MAC 28

1.7.2 Interleaved carrier sense multiple access 29

1.7.3 Two-phase TDMA-based medium access control scheme 31

1.8 Routing protocols for multiradio wireless mesh networks 34

1.8.1 New routing metrics for multiradio wireless mesh networks 34

1.8.2 Multiradio link quality source routing 36

1.8.3 Load-aware interference balanced routing protocol 40

1.9 Topology control schemes for multiradio wireless mesh networks 41

1.9.1 Objectives of topology control protocols 41

1.9.2 The backbone topology synthesis algorithm 42

1.10 Open issues 45

1.11 Summary 46

References 46

1.1 INTRODUCTION Wireless mesh network (WMN) is a radical network form of the ever-evolving wireless networks that marks the divergence from the trad-itional centralized wireless systems such as cellular networks and wireless local area networks (LANs) Similar to the paradigm shift, experienced in wired networks during the late 1960s and early 1970s that led to a hugely successful and distributed wired network form— the Internet—WMNs are promising directions in the future of wireless networks The primary advantages of a WMN lie in its inherent fault tolerance against network failures, simplicity of setting up a network, and the broadband capability Unlike cellular networks

where the failure of a single base station (BS) leading to unavailability

of communication services over a large geographical area, WMNsprovide high fault tolerance even when a number of nodes fail.Although by definition a WMN is any wireless network having anetwork topology of either a partial or full mesh topology, practicalWMNs are characterized by static wireless relay nodes providing adistributed infrastructure for mobile client nodes over a partial meshtopology Due to the presence of partial mesh topology, a WMNutilize multihop relaying similar to an ad hoc wireless network.Although ad hoc wireless networks are similar to WMNs, the proto-cols and architectures designed for the ad hoc wireless networksperform very poorly when applied in the WMNs In addition, theoptimal design criteria are different for both these networks Thesedesign differences are primarily originated from the application

or deployment objectives and the resource constraints in these works For example, an ad hoc wireless network is generally designedfor high mobility multihop environment; on the other hand, a WMN isdesigned for a static or limited mobility environment Therefore, aprotocol designed for ad hoc wireless networks may perform verypoorly in WMNs In addition, WMNs are much more resource-richcompared with ad hoc wireless networks For example, in someWMN applications, the network may have a specific topology andhence protocols and algorithms need to be designed to benefit fromsuch special topologies In addition, factors such as the inefficiency ofprotocols, interference from external sources sharing the spectrum,and the scarcity of electromagnetic spectrum further reduce the cap-acity of a single-radio WMN In order to improve the capacity ofWMNs and for supporting the traffic demands raised by emergingapplications for WMNs, multiradio WMNs (MR-WMNs) are underintense research Therefore, recent advances in WMNs are mainlybased on a multiradio approach While MR-WMNs promise highercapacity compared with single-radio WMNs, they also face severalchallenges This chapter focuses on the issues and challenges for bothsingle-radio WMNs and MR-WMNs, and discusses a set of existingsolutions for MR-WMNs It begins with a comparison of WMNs with

net-ad hoc wireless networks and proceeds to discuss the issues andchallenges in MR-WMNs The main contribution of this chapter is thedetailed discussion on the issues and challenges faced by MR-WMNs,presentation with illustrations of a range of recent solutions for archi-tectures, link layer protocols, medium access control (MAC) layerprotocols, network layer protocols, and topology control solutionsfor MR-WMNs

1.2 COMPARISON BETWEEN WIRELESS

AD HOC AND MESH NETWORKS

Figure 1.1 shows the classification of multihop wireless networks;these constitute the category of wireless networks that primarily usemultihop wireless relaying The major categories in the multihopwireless networks are the ad hoc wireless networks, WMNs, wirelesssensor networks, and hybrid wireless networks This book mainlyfocuses on WMNs Ad hoc wireless networks [12] are mainly infra-structureless networks with highly dynamic topology Wireless sensornetworks, formed by tiny sensor nodes that can gather physicalparameters and transmit to a central monitoring node, can use eithersingle-hop wireless communication or a multihop wireless relaying.Hybrid wireless networks [12] utilize both single- and multihop com-munications simultaneously within the traditionally single-hop wire-less networks such as cellular networks and wireless in local loops(WiLL) WMNs use multihop wireless relaying over a partial meshtopology for its communication

Table 1.1 compares the wireless ad hoc networks and WMNs Theprimary differences between these two types of networks are mobility

of nodes and network topology Wireless ad hoc networks are highmobility networks where the network topology changes dynamically

On the other hand, WMNs do have a relatively static network withmost relay nodes fixed Therefore, the network mobility of WMNs

is very low in comparison with wireless ad hoc networks The

Wireless Ad hoc Networks

Wireless Mesh Networks Multihop Wireless Networks

Wireless Sensor Networks Hybrid Wireless Networks

Figure 1.1 Classification of multihop wireless networks.

topological difference in these networks also contributes to the ence in performance in routing For example, while the on-demandrouting protocols perform better in wireless ad hoc networks, therelatively static hierarchical or table-driven routing protocols performbetter in WMNs Due to the static topology, formed by fixed relaynodes, of WMNs, most WMNs have better energy storage and powersource, thus removing one of the biggest constraint in wireless ad hocnetworks—the energy constraint Finally, another important differ-ence between these two categories of networks is the applicationscenario Unlike wireless ad hoc networks, WMNs are used for bothmilitary and civilian applications Some of the popular civilian appli-cations of WMNs include provisioning of low-cost Internet services toshopping malls, streets, and cities.

differ-Table 1.1 Differences between Ad Hoc Wireless Networks

and Wireless Mesh Networks

Issue

Wireless AdHoc Networks

WirelessMesh Networks

Application

characteristics

permanentInfrastructure

Relaying by fixed nodes

on-demand routingpreferred

Fully distributed orpartially distributedwith table-driven orhierarchical routingpreferred

required

sensor trafficPopular application

scenario

Tacticalcommunication

Tactical and civiliancommunication

1.3 CHALLENGES IN WIRELESS MESH NETWORKSTraditional wireless ad hoc networks and WMNs were based on asingle-channel or single-radio interface WMNs, irrespective of itssimplicity and high fault tolerance, face a significant limitation oflimited network capacity While the theoretical upper limit of the

n log n) where n is thenumber of nodes in the network Therefore, with increasing num-ber of nodes in a network, the throughput capacity becomesunacceptably low With the use of real MAC, routing, and transportprotocols and a realistic traffic pattern, the achievable capacity in

a WMN, in practice, is much less than the theoretical upper limit

It has also been found [7] through experiments using carrier sensemultiple access with collision avoidance (CSMA/CA)-based MACprotocol such as IEEE 802.11 that on a string topology, thethroughput degrades approximately to 1/n of the raw channelbandwidth In general, the throughput capacity achievable in an

dimension of the network and W is the total bandwidth For a dimensional (2D) network, the throughput can be as small as

of a WMN is to use multiple radio interfaces Although the upperlimit of the capacity is unaffected by the raw bandwidth or theway the raw bandwidth is split among multiple interfaces, inpractice, with realistic MAC and routing protocols, the throughputcapacity can be significantly increased by the use of multiple inter-faces and by fine tuning of protocols Recently, the develop-ment of WMNs using multiple radio interfaces have takensignificant process due to the availability of inexpensive and off-the-shelf IEEE 802.11-based wireless interfaces While MR-WMNsprovide several advantages such as increased network capacity,they also face several issues and challenges This chapter primarilyfocuses on the issues and challenges in single-radio and MR-WMNsand proceeds to discuss some of the solutions for a multiradiowireless network The challenges faced by WMNs are discussed inSection 1.3.1 through Section 1.3.4

The primary challenges faced by WMNs such as throughput acity, network scalability, and other challenges are discussed here

cap-1.3.1 Throughput Capacity

The throughput capacity achievable for WMN nodes is limited in asingle-channel system compared to a multichannel system Table 1.2shows the throughput deviation in a string topology, as depicted inFigure 1.2, over one, two, and three hops, in a typical experimentalnetwork From Table 1.2, it can easily be found that throughputdegrades rapidly with a WMN system as the path length increases.Although there are several factors contributing to the throughputdegradation, such as characteristics of MAC protocol, the exposednode problem, the hidden terminal problem, and the unpredictableand high error rate in the wireless channel, all these issues are aggra-vated in a single-channel system For example, as illustrated in Figure1.2, when node 1 transmits to node 2, especially when CSMA/CA-based MAC protocols are employed, nodes 2 and 3 cannot initiateanother transmission Node 2 is prevented from a simultaneous trans-mission as the wireless interface, in most WMNs is half-duplexwhereas node 2 abstains from transmission because it is exposed tothe ongoing transmission between nodes 1 and 2 This exposed nodeproblem contributes to the throughput degradation in WMNs over arelayed multihop path For example, a two-hop flow between nodes 1and 3 has to share the bandwidth between the two and therefore, fromTable 1.2, the end-to-end throughput for a two-hop path is only 47%

of the single-hop throughput In experimental arbitrary dimensional (ID) networks, the throughput degradation is found to

one-be following a function of O(1/n) where n is the numone-ber of hops

Table 1.2 Throughput Degradation in a WMN with String Topology

1 Hop 2 Hops 3 Hops 4 Hops 5 Hops >5 Hops

1

Figure 1.2 An example of string topology and exposed node problem in

a wireless mesh network.

when the hop length is less than five hops and beyond five hops, thethroughput remains constant albeit at a very low value.

Although there exist other factors such as the nature of routingprotocol, greediness of the initial nodes and subsequent flow starva-tion of the latter hops, and the behavior of MAC protocols, the singlemost important factor contributing such a rapid degradation ofthroughput is the exposed node problem, aggravated by the use of asingle-radio system

1.3.2 Throughput Fairness

Another important issue in a single-radio WMN is the high throughputunfairness faced by the nodes in the system A network is said to beexhibiting high throughput fairness if all nodes get equal throughputunder similar situations of source traffic and network load WMNsshow high throughput unfairness among the contending traffic flowsespecially when CSMA/CA-based MAC protocols are employed forcontention resolution Figure 1.3a and Figure 1.3b show simple topol-ogies within a WMN, causing high throughput unfairness

Two important properties associated with CSMA/CA-based MACprotocols, when used in a WMN environment are: (i) informationasymmetry depicted in Figure 1.3a, (ii) location-dependent contentiondepicted in Figure 1.3b, and (iii) half-duplex character of single-channel systems In Figure 1.3a, only the the receiver of the trafficflow P is exposed to both the sender and the receiver of flow Q, andtherefore, the sender of the flow P does not get any information fromthe channel about ongoing transmissions on other flows On the other

Coverage of traffic flow P

Coverage of traffic flow Q

P 2

(b) (a)

Figure 1.3 Traffic flows and throughput unfairness in WMNs.

hand, the channel activity is known to be the sender of flow Q Thisinformation asymmetry causes unfair sharing of the total throughputachieved That is, among the flows P and Q, it is seen [1] that the flow

P receives about 5% of the total throughput compared with the 95%throughput achieved by flow Q For example, when node 1 haspackets ready for transmission, upon detecting an idle channel, itmay start transmission by sending request-to-send (RTS) packet tonode 2 At this point, if there is an ongoing transmission betweennodes 3 and 4, node 2 does not respond to the RTS, leaving node 1 toexponentially back off and retry again This repeated back-off andseveral retransmission attempts lead to achieving a low throughput forflow Q On the other hand, a similar situation can happen to node 3with a much lower probability and that is proportional to the vulner-able period of the medium access scheme, which in this case is thepropagation delay between nodes 2 and 3

While the information asymmetry is caused by lack of information

at certain nodes, having excessive information may also contribute tothroughput unfairness For example, in Figure 1.3b, flows P and R donot have information about any other flows in the network whereasflow Q has information about both the other flows Therefore, flow Qhas to set its network allocation vector (NAV) and abstain from trans-mitting, whenever it sees a transmission of control packets or datapackets belonging to flows P and R This leads the flow Q to wait for

an idle channel that essentially depends on the event of both the flows

P and Q simultaneously going idle In this case, the location of theflow Q is in such a position that it experiences much more contentionthan the rest of the flows [1] and therefore, flow Q receives only 28% ofthe total throughput compared with 36% throughput share received

by both the flows P and R In fact, the throughput share of flow Q isinversely proportional to the number of neighbor flows contendingfor its bandwidth Another effect of this location-dependent conten-tion is known as perceived collision, which may occur at flow Q Due

to the presence of contending flows, trying to access the channel,simultaneous transmission of control packets, RTS, CTS, and ACK byboth the flows P and R, may result in a collision at flow Q This results

in a wrong perception of collision at flow Q that in fact may not be acollision for both the flows P and R This perceived collision mayreduce the amount of information, about the flows P and R, available

at the flow Q and, therefore, leading to further degradation ofthroughput fairness

In addition to the information asymmetry and the location-dependentcontention, the half-duplex property of a single-interface system is

another property that causes high throughput unfairness in a radio WMN Due to the half-duplex characteristics, no node cansimultaneously receive and transmit This is illustrated in Figure 1.4a

single-in which a ssingle-ingle half-duplex radio with a channel data rate of B bits/s

is employed and therefore, only one communication could be ted at any time Therefore, only flow P could be transmitting whileother nodes are waiting As mentioned earlier, in certain MAC proto-cols such as CSMA/CA-based IEEE 802.11, there exists a strong chance

permit-of channel capture where a successful node keeps getting sion opportunities more often than others Such channel capturingand subsequent unfairness can be prevented by using multiple chan-nels In Figure 1.4b, each node uses two radio interfaces with channeldata rate of B/2 bits/s and therefore, two simultaneous flows, P and Q,could exist In this case, though each channel has only half thebandwidth, the throughput fairness increases as found in experimen-tal studies [8]

transmis-1.3.3 Reliability and Robustness

Another important motivation for using WMNs and especially theMR-WMNs is to improve the reliability and robustness of communica-tion The partial mesh topology in a WMN provides high reliabilityand path diversity against node and link failures MR-WMNs providethe most important ingredient for robustness in communication—diversity For example, in wireless systems channel errors can be veryhigh compared to wired networks; therefore, graceful degradation of

Coverage of traffic flow P

Coverage of traffic flow Q

P 2

1

3 4

P 2

1

3Q 4

Flow over a link with bandwidth B

(b) (a)

Flow over a link with bandwidth B/2

Figure 1.4 Half-duplex radio interfaces in WMNs.

communication quality during high channel errors is necessary This isparticularly important when the WMN system utilizes unlicensed fre-quency spectrum [9] In order to achieve graceful quality degradationinstead of full loss of connectivity, WMNs can employ frequency diver-sity, by using multiple radio interfaces, which is difficult to achieveinasingle-radio WMN system MR-WMNs can use appropriate radio-switching modules to achieve fault tolerance in communicationeither by switching the radios, channels, or by using multiple radiossimultaneously.

1.3.4 Resource Management

Resource management refers to the efficient management of networkresources such as energy, bandwidth, interfaces, and storage Forexample, the energy resources can be efficiently used in a WMNwith limited energy reserve if each node in the system has a newlow-power interface in addition to the regular interface The overallpower consumption, even in idle mode, depends very much on thetype of interface Therefore, in an IEEE 802.11-based WMN withlimited energy reserve, an additional low-power and low-data rateinterface can be used to carry out-of-band signaling information tocontrol the high-power and high-data rate data interface Bandwidthresources can also be managed better in a multiradio environment.For example, the load balancing across multiple interfaces could helppreventing any particular channel getting heavily congested andhence becoming a bottleneck In addition to balancing the load,bandwidth achieved through each interface can be aggregated toobtain a high effective data rate In such a bandwidth aggregationmechanism (also known as bandwidth striping), dynamic packetscheduling can be utilized to obtain a better performance Finally,one important advantage of using a multiradio system in a WMN isthe possibility to effect provisioning quality of service through servicedifferentiation

1.4 DESIGN ISSUES IN WIRELESS MESH NETWORKS

There are many issues that need consideration when a WMN isdesigned for a particular application These design issues can bebroadly classified into architectural issues and protocol issues Thearchitectural design issues and protocol design issues are described inSection 1.4.1 and Section 1.4.2

1.4.1 Network Architectural Design Issues

A WMN can be designed in three different network architectures based

on the network topology: flat WMN, hierarchical WMN, and hybridWMN These categories are briefly discussed below

1.4.1.1 Flat Wireless Mesh Network

In a flat WMN, the network is formed by client machines that act asboth hosts and routers Here, each node is at the same level as that ofits peers The wireless client nodes coordinate among themselves toprovide routing, network configuration, service provisioning, andother application provisioning This architecture is closest to an

ad hoc wireless network and it is the simplest case among the threeWMN architectures The primary advantage of this architecture is itssimplicity, and its disadvantages include lack of network scalabilityand high resource constraints The primary issues in designing a flatWMN are the addressing scheme, routing, and service discoveryschemes In a flat network, the addressing is one of the issues thatmight become a bottleneck against scalability

1.4.1.2 Hierarchical Wireless Mesh Network

In a hierarchical WMN, the network has multiple tiers or hierarchicallevels in which the WMN client nodes form the lowest in the hier-archy These client nodes can communicate with a WMN backbonenetwork formed by WMN routers In most cases, the WMN nodes arededicated nodes that form a WMN backbone network This means thatthe backbone nodes may not originate or terminate data traffic like theWMN client nodes The responsibility to self-organize and maintainthe backbone network is provided to the WMN routers, some of which

in the backbone network may have external interface to the Internetand such nodes are called gateway nodes

1.4.1.3 Hybrid Wireless Mesh Network

This is a special case of hierarchical WMNs where the WMN utilizesother wireless networks for communication For example, the use ofother infrastructure-based WMNs such as cellular networks, WiLL net-works, WiMAX networks, or satellite networks Examples of suchhybrid WMNs include multihop cellular networks [2], throughputenhanced wireless in local loop (TWiLL) networks [3], and unifiedcellular ad hoc networks [4] A practical solution for such a hybridWMN for emergency response applications is the CalMesh platform[5] These hybrid WMNs may use multiple technologies for both WMN

backbone and back haul Since the growth of WMNs depend heavily onhow it works with other existing wireless networking solutions, thisarchitecture becomes very important in the development of WMNs.

1.4.2 Network Protocol Design Issues

The design issues for the protocols can be described in a layer-wisemanner starting from the physical layer to the application layer Some

of these protocol design issues are presented below

1.4.2.1 Physical Layer Design Issues

At the physical layer, the main design issue is the choice of anappropriate radio technology The choice of a radio technology can

be based on: (i) technological considerations and (ii) economic siderations The main technological considerations include the spec-tral efficiency, physical layer data rate, and the ability to operate in thepresence of interference For example, the choice of technologiessuch as code division multiple access (CDMA), ultra wide band(UWB), and multiple input multiple output (MIMO) are more suitablefor WMN physical layer than the most popular physical layer technol-ogy, orthogonal frequency division multiplexing (OFDM) used intoday’s WMNs For example, today’s physical layer technology, pri-marily based on OFDM provides a maximum physical layer data rate

con-of 54 Mbps In a highly dense network with high interference, thiscapacity may not be sufficient Therefore, development of new andhigh data rate physical layer such as UWB is a physical layer challenge

In addition to the choice of a particular physical layer technology,programable radios or cognitive radios add another dimension to theWMN physical layer design This is emphasized by some of theapplications of WMNs such as emergency response and militaryapplications where the spectrum used for communication depends

on the unused spectrum in a given locality In such applications, asoftware-defined radio with cognitive capabilities would be an idealchoice In addition to the technological considerations mentionedabove, the second most important requirement is economical or socialwhere the simplicity of the physical layer technology will lead toinexpensive devices and hence better social affordability of WMNs

An example of this is evident in the success of today’s IEEE based WMNs where the inexpensive network interface cards contrib-uted to the success of the proliferation of WMNs Therefore, whilechoosing the physical layer technology, a network designer shouldlook at the application and user scenario as well

802.11b-1.4.2.2 Medium Access Control Layer

The design of MAC layer protocol assumes significance in a WMNbecause achievable capacity depends heavily on the performance ofMAC protocol In addition to a fully distributed operation, the majorissues faced by the popular CSMA/CA-based IEEE 802.11 distributedcoordination function (DCF) are: (i) hidden terminal problem, (ii)exposed terminal problem, (iii) location-dependent contention, and(iv) high error probability on the channel In order to increase thenetwork capacity, multiple radios operating in multiple channels areused Therefore, new MAC protocols are to be designed for operating

in multichannel MR-WMN systems MAC protocols are also to beadapted to operate in different physical layer technologies such asUWB and MIMO physical layers Another popular research issue forbetter MAC performance is the use of cross-layer interaction mechan-isms that enable the MAC protocol to make use of information fromother layers In traditional wireless or wired networks, each layerworks with its own information making it unable to make the bestuse of the network-centric properties In general, the MAC layerprotocol design should include methods and solutions to providebetter network scalability and throughput capacity

1.4.2.3 Network Layer

Unlike the routing protocols for ad hoc wireless networks, the routingprotocols, depending on its network scenario, face different designissues in a WMN Since WMN is relatively a static network, the routingcan make use of table-driven routing approaches such as that used inwired networks or in ad hoc wireless networks [12] The main issuesfaced by routing protocol in a WMN are: (i) design of routing metric,(ii) minimal routing overhead, (iii) route robustness, (iv) effective use

of support infrastructure, (v) load balancing, and (vi) route ity The routing metric design plays a crucial role in achieving goodperformance The best routing metric may also differ in its perform-ance For example, in WMNs, the routing metric design has to take thelink level signal quality into account for better end-to-end perform-ance Routing protocols for WMNs, while providing a good end-to-endperformance, should also consume minimum bandwidth for setting

adaptabil-up paths In addition, the use of wireless medium demands quickpath reconfiguration capability in order to maintain the robustness

of the path Another important aspect is the load-balancing ability that needs to be incorporated with the routing protocol Finally,

cap-a routing protocol for WMN must be cap-adcap-aptcap-able to the network

dynamics The routing protocol can be classified into either flat routingprotocol or hierarchical routing protocol based on the type of networkwhere the routing protocol is applied.

1.4.2.4 Transport Layer

At the transport layer, the biggest challenge is the performance oftransport protocols over the WMN Since a WMN has large round-triptime (RTT) variations and these RTT variations are dependent on thenumber of hops in the path, the end-to-end TCP throughput degradesrapidly with throughput The packet loss, collision, network asymmetry,and link failures can also contribute to the degradation in transport layerprotocol performance The popular transport layer for the Internet, TCP,performs very poorly in its original form over a WMN The transportlayer needs to be refined or rewritten for making it more efficient on aWMN Some of the design issues for a transport layer protocol for WMNare: (i) end-to-end reliability, (ii) throughput, (iii) capability to handlenetwork asymmetry, and (iv) capability to handle network dynamism

1.4.2.5 Application Layer

The most popular application for WMNs is the Internet access service.Essentially, a WMN needs to provide Internet services for residentialareas or businesses In such a situation, though data services makeprimary service over a WMN, voice services such as voice over Inter-net protocol (VoIP) are also important Therefore, it is very essential toprovide support for both the time-sensitive and the best-effort traffics

In addition to the basic data and voice traffic support, the networkprovides service discovery mechanisms Since most of the networkservices are in fully distributed form, static service discovery mechan-isms may not be effective in a WMN Another important requirementfor the application layer protocol design is to handle the heterogeneity

of networks as the data may pass through a variety of networks beforebeing delivered to the end application

1.4.2.6 System-Level Design Issues

The above-mentioned issues are generic to a WMN and these issuesare revisited in detail for a MR-WMN system in Section 1.5 In addition

to the protocol design issues, a WMN requires system-level solutions.Some examples for system-level issues are: (i) cross-layer systemdesign, (ii) design for security and trust, (iii) network managementsystems, and (iv) network survivability issues

Some of the primary challenges faced by a WMN can be alleviated bythe use of an MR-WMN and therefore, subsequent sections focus on this

1.5 DESIGN ISSUES IN MULTIRADIO WIRELESS

MESH NETWORKSThe primary advantages of using an MR-WMN are the improvedcapacity, scalability, reliability, robustness, and architectural flexibility.Notwithstanding the advantages of using a multiradio system forWMNs, there exist many challenges for designing an efficientMR-WMN system This section discusses the issues to be considered fordesigning an MR-WMN The main issues can be classified into archi-tectural design issues, MAC design issues, routing protocol designissues, and routing metric design issues, which are explained below

1.5.1 Architectural Design Issues

The network architecture plays a major role in achieving the ance objectives of an MR-WMN when a network is deployed Ingeneral, the network architecture of an MR-WMN is designed on thebasis of the type of application or deployment scenario The majorarchitectural choices to be considered are: (a) topology-based, (b)technology-based, and (c) node-based Based on the topology, anMR-WMN can be designed either as a flat-topology-based or as ahierarchical-topology-based The design categories under the technol-ogy-based solution are homogeneous or heterogeneous While themost popular form of MR-WMN system is the homogeneous type thatuses only one type of radio technology such as the popular WLANtechnology IEEE 802.11, it is possible to develop an MR-WMN withheterogeneous technologies that utilize a variety of communicationtechnologies Finally, the node-based design criteria can be classifiedinto either host-based, infrastructure-based, or hybrid MR-WMNs Inthe case of host-based MR-WMNs, the network is formed by the hostnodes and is same as an ad hoc wireless network with limited or nomobility On the other hand, in the infrastructure-based MR-WMNs,the WMN is formed by nodes placed on fixed infrastructures orbuildings An example for this architectural type is the rooftop net-works formed by placing wireless mesh relay nodes on the roof ofevery house for building a residential communication network.Finally, a hybrid MR-WMN has both infrastructure-based backboneand wireless mesh hosts These hosts communicate over the wirelessmesh backbone This backbone topology can be organized either as aflat topology or as a hierarchical topology as discussed in Section 1.4.1

perform-In some application environments, the hosts are mobile and they alsorelay traffic on behalf of other hosts in the network An example of suchhybrid MR-WMNs is the vehicular WMNs that communicate over

a wireless mesh infrastructure Therefore, the design of an MR-WMNsystem must consider the type of application or deployment environ-ment for choosing appropriate architectural solution.

1.5.2 Medium Access Control Design Issues

The MAC layer for MR-WMNs faces several challenges The mainchallenges among them are the interchannel interference, interradiointerference, channel allocation, and MAC protocol design The inter-channel interference refers to the interference experienced at a givenchannel due to the activity in neighbor channels For example, in IEEE802.11b, although there are a total of 11 unlicensed channels in NorthAmerica (13 in Europe and 14 in Japan), only 3 of them (channels 1, 6,and 11 in North America) can be used simultaneously at any givengeographical location Therefore, the presence of multiple radios mustconsider the interchannel interference as the use of a new channel, at

a second interface that interferes with the existing channel, will lead tosignificant performance degradation In such cases, multiradio chan-nel usage must use nonoverlapping channels The second issue here

is the interradio interference This issue arises due to the design andimplementation of radio interfaces This type of interference is experi-enced at a particular radio due to the channel activity at anotherinterface in the same WMN node Such interferences occur evenwhen both the interfaces use nonoverlapping channels [27] Forexample, when interfaces A and B on a WMN node use channels 1and 11, respectively, interradio interference may experience Thisinterference is primarily due to the design of the hardware compon-ents and the interface itself where usually a number of low-cost filtersand associated RF components are used The physical separation ofinterfaces may help to avoid this issue to some extent; in certain casesthe separation may be difficult, especially in portable nodes The use

of certain low-cost interface cards leads to interference even whenthey are separated for few feet [27] Another issue of importance toMAC is the channel allocation This is a network-wide process wherethe allocation of noninterfering channels would lead to significantthroughput and media access performance The channel allocationshould consider the number of channels available and the number ofinterfaces available Therefore, techniques such as graph coloring areused for generating channel allocation strategies Finally, the mostimportant issue is the design of MAC protocols The availability ofmultiple interfaces and multiple channels leads to new designs formedium access protocols that can be benefitted in the presence of

multiple radios Examples of such protocols are the multichannelcarrier sense multiple access (MCSMA) [24], interleaved carrier sensemultiple access (ICSMA) [8], and the two-phase time division multipleaccess (2P-TDMA) [26] These protocols utilize multiple channelssimultaneously and also attempt to solve the media access issue inMR-WMNs.

1.5.3 Routing Protocol Design Issues

Another important issue in designing an MR-WMN is the design ofrouting protocol that depends on the design of the WMN architectureand in some cases it also depends on both the network’s applicationand the deployment scenario The routing protocol design can beclassified into several categories based on: (a) the routing topology,(b) the use of a routing backbone, and (c) the routing informationmaintenance approach Based on the routing topology, routing proto-cols can be designed either as a flat routing protocol or as a hierarch-ical routing protocol In hierarchical routing, a routing hierarchy isbuilt among the nodes in such a way that the pathfinding responsibil-ity is delegated to higher-level nodes in the hierarchy when the lowerlevel nodes fail to obtain a path, e.g., hierarchical state routing (HSR)[11] On the other hand, a flat routing system does not have any inbuilthierarchies and each node has equal responsibility to find a path to thedestination and to participate in the pathfinding process of othernodes The chosen path may include any arbitrary node in the net-work without following any particular node hierarchy Second designcategory is routing based on routing backbones and is classified intotree-based backbone routing, mesh-based backboneless routing, andhybrid topology routing Unlike a wireless ad hoc network, a WMN isrelatively static or has limited mobility network; therefore, in order toincrease the routing efficiency, a routing backbone can be built Anexample of this routing approach is the WMN routing performed bythe IP routing mechanism over a spanning tree protocol (STP)-basedtree backbone [10] In the case of STP, the link layer will form a treetopology among the WMN nodes similar to a wireless distributedsystem (WDS) [12] and at the network layer, routing is carried out

by traditional IP-based routing method Although this is one of thesimplest approach for WMNs, it has several issues such as poorreliability and lack of network scalability On the other hand, routingprotocols designed for and implemented at the network layer mayfollow a backboneless mesh routing approach A third approach is

to use a network layer backbone-topology, a subset of the nodes

forming a mesh-like backbone within the WMN, optimized for certainparameters such as throughput, channel quality, or network scalabilitycan be used for aiding a backboneless mesh routing protocol Such arouting approach that uses a dynamic backbone topology at certainspecific segment of the network is called a hybrid topology routingprotocol Finally, routing protocols can be designed on the basis of therouting information maintenance approach Examples of such routingschemes are proactive or table-driven routing protocols, reactive oron-demand routing protocols, and hybrid routing protocols In thecase of proactive or table-driven routing approach, every nodeexchanges its routing information periodically and maintains a routingtable, which contains routing information to reach every node in thenetwork Examples of routing protocols that use this design approachare DSDV [13], WRP [14], and STAR [15] On the other hand, in thereactive or on-demand routing approach, a node requests routinginformation and maintains the path information only when it needs

to communicate with another node Some of the routing protocols,based on this approach, are AODV [16], dynamic source routing (DSR)[17], and multiradio link quality source routing (MRLQSR) [20] Finally,the hybrid routing protocols take benefit of both the table-driven andon-demand routing approaches An example of such a hybrid routingapproach is the zone routing protocol (ZRP) [18], which employs atable-driven routing approach within a zone and on-demand approachbeyond the zone That is, every node uses proactive approach within ak-hop routing zone and employs a reactive routing approach beyondthe routing zone

1.5.4 Routing Metric Design Issues

In addition to designing a routing protocol, another important issue isthe design of a routing metric A routing metric is the routing param-eter, weight, or value that is associated with a link or path, based onwhich a routing decision is made Hop count is the simplest routingmetric and is an additive routing metric Due to the special character-istics of WMNs, hop count as a routing metric performs very poorly.Therefore, the design of routing metric is very important in MR-WMNs.The routing metric plays a crucial role in the performance of a routingprotocol and the design of routing metrics should take several factorssuch as (i) the network architecture, (ii) the network environment, (iii)the extent of network dynamism, and (iv) the basic characteristics ofthe routing protocol into account, in order to design an efficientrouting protocol for WMNs First, the architectural property of the

network needs to be considered for designing the routing metric Forexample, the network may be designed based on either a flat topologyarchitecture or a multitiered hierarchical architecture In addition,the architectural design may include an infrastructureless network,partially infrastructure-supported network, or an infrastructure-supported network The routing metric to be designed should takethe architectural design of the network into account Second import-ant factor to be considered is the network environment For example,due to the presence of location-dependent contention, highly fluctu-ating and unpredictable channel conditions, and high bit error rate(BER), the characteristics of a WMN environment is radically differentfrom that of a wired network Therefore, the design of routing proto-col and routing metric should take the specific network environmentinto account Another important input for designing a WMN is theextent of the network dynamism due to the mobility experienced bythe network For a WMN designed for static nodes or for nodes withvery low mobility, proactive protocols may be suitable whereas on-demand routing approach is suitable for a WMN that handles highmobility nodes Finally, in order to design an efficient routing metric,the basic characteristics of a routing approach is important Forexample, while a nonisotonic* routing protocol works well with anon-demand source-routing-based routing protocol, it may fail or per-form poorly due to the formation of routing loops when used with atable-driven hop-by-hop routing protocol.

The design objectives for a routing protocol and metric are: (i)resource efficiency, (ii) throughput, (iii) freedom from routing loops,(iv) route stability, (v) quick path setup capability, and (vi) efficientroute maintenance

The challenges for designing routing protocols for MR-WMNs are:(a) interradio interference, (b) interflow interference, (c) intraflowinterference, (d) hidden terminal node problem, (e) exposed terminalnode problem, (f) location-dependent contention, and (g) highlydynamic channel characteristics

1.5.5 Topology Control Design Issues

The network performance in a WMN is affected by the network topology,and by controlling the network topology the network performance can

* Isotonicity is the property of a routing metric that guarantees freedom from routing loops.

be improved Topology control is defined by the network’s capability tomanipulate its parameters such as the location of nodes, mobility of nodes,transmission power, the properties of the antenna, and the status of thenetwork interfaces The topology can be controlled either as a one-timeactivity during the network initialization phase or as a periodic activitythroughout the duration of the network lifetime Effective use of thenetwork topology control can help to improve the capacity In practice,node location and mobility are not under the direct control of the networksystem leaving the remaining factors such as transmission power, antennaproperties, and the status of the network interface cards The objectives oftopology control mechanisms are connectivity, capacity, reliability andfault tolerance, and network coverage Section 1.9 provides a detaileddescription of the objectives of topology control.

1.6 LINK LAYER SOLUTIONS FOR MULTIRADIO

WIRELESS MESH NETWORKSNetwork scalability is the single most important problem that plaguesthe large-scale WMNs The primary reasons behind the lack of net-work scalability in a WMN are: (i) half-duplex character of the WLANradios, (ii) inefficient interaction between the network congestion andsuboptimal congestion avoidance phase at different layers of theprotocol stack, (iii) collision due to hidden terminal problem, (iv)resource wasted due to exposed terminal problem and the location-dependent contention, and (v) the difficulties in handling a multi-channel system Some of the above-mentioned problems can besolved by an MR-WMN However, they face several challenges such

as (i) adjacent radio interference, (ii) dynamic management ofspectrum resources, and (iii) efficient management of multiple radiointerfaces The adjacent radio interface problem refers to the interfer-ence caused by one radio interface to the other radio interfaces on thesame node The only way to mitigate this problem is to modify themechanical design of the nodes to provide enough separation ofthe antennas or network interface cards The dynamic spectrummanagement can be provided by intelligently choosing the mostappropriate channel for communication between nodes The man-agement of multiple radio interfaces refers to the activity by which thehigher layer protocols could use the desired radio interface that isappropriate for communication There exist several link layer solu-tions such as multiradio unification protocol (MUP) that is discussed

in detail in Section 1.6.1