Lecture notes on Wireless Mesh Networks

WIRELESS MESH NETWORKING Architectures, Protocols and Standards how to implement wireless mesh network how to configure wireless mesh network. and wireless mesh network advantages and disadvantages pdf free download
VoiletFord Profile Pic
VoiletFord,United States,Professional
Published Date:12-07-2017
Your Website URL(Optional)
Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 3 23.10.2006 1:43pm 1 WIRELESS MESH NETWORKS: ISSUES AND SOLUTIONS B.S. Manoj and Ramesh R. Rao CONTENTS 1.1 Introduction ........................................ ................................................................................ .......................................... 5 4 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 Flat wireless mesh network.................................. 14 Hierarchical wireless mesh network .................... 14 Hybrid wireless mesh network............................. 14 1.4.2 Network protocol design issues .....................................15 Physical layer design issues.................................. 15 Medium access control layer ................................ 16 Network layer........................................................ 16 Transport layer ...................................................... 17 Application layer................................................... 17 System-level design issues.................................... 17 3Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 4 23.10.2006 1:43pm & 4 Wireless Mesh Networking 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 networksYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 5 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 5 where the failure of a single base station (BS) leading to unavailability of communication services over a large geographical area, WMNs provide high fault tolerance even when a number of nodes fail. Although by definition a WMN is any wireless network having a network topology of either a partial or full mesh topology, practical WMNs are characterized by static wireless relay nodes providing a distributed infrastructure for mobile client nodes over a partial mesh topology. Due to the presence of partial mesh topology, a WMN utilize 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 networks perform very poorly when applied in the WMNs. In addition, the optimal design criteria are different for both these networks. These design differences are primarily originated from the application or deployment objectives and the resource constraints in these net- works. For example, an ad hoc wireless network is generally designed for high mobility multihop environment; on the other hand, a WMN is designed for a static or limited mobility environment. Therefore, a protocol designed for ad hoc wireless networks may perform very poorly in WMNs. In addition, WMNs are much more resource-rich compared with ad hoc wireless networks. For example, in some WMN applications, the network may have a specific topology and hence protocols and algorithms need to be designed to benefit from such special topologies. In addition, factors such as the inefficiency of protocols, 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 of WMNs and for supporting the traffic demands raised by emerging applications for WMNs, multiradio WMNs (MR-WMNs) are under intense research. Therefore, recent advances in WMNs are mainly based on a multiradio approach. While MR-WMNs promise higher capacity compared with single-radio WMNs, they also face several challenges. This chapter focuses on the issues and challenges for both single-radio WMNs and MR-WMNs, and discusses a set of existing solutions for MR-WMNs. It begins with a comparison of WMNs with ad hoc wireless networks and proceeds to discuss the issues and challenges in MR-WMNs. The main contribution of this chapter is the detailed 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) layer protocols, network layer protocols, and topology control solutions for MR-WMNs.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 6 23.10.2006 1:43pm & 6 Wireless Mesh Networking 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 use multihop wireless relaying. The major categories in the multihop wireless networks are the ad hoc wireless networks, WMNs, wireless sensor networks, and hybrid wireless networks. This book mainly focuses on WMNs. Ad hoc wireless networks 12 are mainly infra- structureless networks with highly dynamic topology. Wireless sensor networks, formed by tiny sensor nodes that can gather physical parameters and transmit to a central monitoring node, can use either single-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 mesh topology for its communication. Table 1.1 compares the wireless ad hoc networks and WMNs. The primary differences between these two types of networks are mobility of nodes and network topology. Wireless ad hoc networks are high mobility networks where the network topology changes dynamically. On the other hand, WMNs do have a relatively static network with most relay nodes fixed. Therefore, the network mobility of WMNs is very low in comparison with wireless ad hoc networks. The Multihop Wireless Networks Hybrid Wireless Networks Wireless Mesh Wireless Ad hoc Networks Networks Wireless Sensor Networks Figure 1.1 Classification of multihop wireless networks.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 7 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 7 Table 1.1 Differences between Ad Hoc Wireless Networks and Wireless Mesh Networks Wireless Ad Wireless Issue Hoc Networks Mesh Networks Network topology Highly dynamic Relatively static Mobility of relay nodes Medium to high Low Energy constraint High Low Application Temporary Semipermanent or characteristics permanent Infrastructure Infrastructureless Partial or fully fixed requirement infrastructure Relaying Relaying by mobile Relaying by fixed nodes nodes Routing performance Fully distributed Fully distributed or on-demand routing partially distributed preferred with table-driven or hierarchical routing preferred Deployment Easy to deploy Some planning required Traffic characteristics Typically user traffic Typically user and sensor traffic Popular application Tactical Tactical and civilian scenario communication communication topological difference in these networks also contributes to the differ- ence in performance in routing. For example, while the on-demand routing protocols perform better in wireless ad hoc networks, the relatively static hierarchical or table-driven routing protocols perform better in WMNs. Due to the static topology, formed by fixed relay nodes, of WMNs, most WMNs have better energy storage and power source, thus removing one of the biggest constraint in wireless ad hoc networks—the energy constraint. Finally, another important differ- ence between these two categories of networks is the application scenario. Unlike wireless ad hoc networks, WMNs are used for both military and civilian applications. Some of the popular civilian appli- cations of WMNs include provisioning of low-cost Internet services to shopping malls, streets, and cities.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 8 23.10.2006 1:43pm & 8 Wireless Mesh Networking 1.3 CHALLENGES IN WIRELESS MESH NETWORKS Traditional wireless ad hoc networks and WMNs were based on a single-channel or single-radio interface. WMNs, irrespective of its simplicity and high fault tolerance, face a significant limitation of limited network capacity. While the theoretical upper limit of the per node throughput capacity is asymptotically limited by pffiffiffi O(1= n), theoretically achievable capacity to every node in a random static wireless ad hoc network, with ideal global schedul- p ing and routing, is estimated 6 as Q(1/ n log n)where n is the number of nodes in the network. Therefore, with increasing num- ber of nodes in a network, the throughput capacity becomes unacceptably low. With the use of real MAC, routing, and transport protocols 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 sense multiple access with collision avoidance (CSMA/CA)-based MAC protocol such as IEEE 802.11 that on a string topology, the throughput degrades approximately to 1/n of the raw channel bandwidth. In general, the throughput capacity achievable in an 1=d arbitrary WMN is proportional to theQ(W n )where d is the dimension of the network and W is the total bandwidth. For a two- dimensional (2D) network, the throughput can be as small as 1=2 Q(W n ). One approach to improve the throughput capacity of a WMN is to use multiple radio interfaces. Although the upper limit of the capacity is unaffected by the raw bandwidth or the way the raw bandwidth is split among multiple interfaces, in practice, with realistic MAC and routing protocols, the throughput capacity 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 taken significant process due to the availability of inexpensive and off- the-shelf IEEE 802.11-based wireless interfaces. While MR-WMNs provide several advantages such as increased network capacity, they also face several issues and challenges. This chapter primarily focuses on the issues and challenges in single-radio and MR-WMNs andproceedstodiscuss some of the solutions for a multiradio wireless network. The challenges faced by WMNs are discussed in Section 1.3.1 through Section 1.3.4. The primary challenges faced by WMNs such as throughput cap- acity, network scalability, and other challenges are discussed here.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 9 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 9 Table 1.2 Throughput Degradation in a WMN with String Topology 1 Hop 2 Hops 3 Hops 4 Hops 5 Hops 5 Hops Normalized throughput 1 0.47 0.32 0.23 0.15 0.14 1 1 0.5 0.33 0.25 0.2 0.16 Hoplength 1.3.1 Throughput Capacity The throughput capacity achievable for WMN nodes is limited in a single-channel system compared to a multichannel system. Table 1.2 shows the throughput deviation in a string topology, as depicted in Figure 1.2, over one, two, and three hops, in a typical experimental network. From Table 1.2, it can easily be found that throughput degrades rapidly with a WMN system as the path length increases. Although there are several factors contributing to the throughput degradation, such as characteristics of MAC protocol, the exposed node problem, the hidden terminal problem, and the unpredictable and high error rate in the wireless channel, all these issues are aggra- vated in a single-channel system. For example, as illustrated in Figure 1.2, when node 1 transmits to node 2, especially when CSMA/ CA-based MAC protocols are employed, nodes 2 and 3 cannot initiate another transmission. Node 2 is prevented from a simultaneous trans- mission as the wireless interface, in most WMNs is half-duplex whereas node 2 abstains from transmission because it is exposed to the ongoing transmission between nodes 1 and 2. This exposed node problem contributes to the throughput degradation in WMNs over a relayed multihop path. For example, a two-hop flow between nodes 1 and 3 has to share the bandwidth between the two and therefore, from Table 1.2, the end-to-end throughput for a two-hop path is only 47% of the single-hop throughput. In experimental arbitrary one- dimensional (ID) networks, the throughput degradation is found to be following a function of O(1/n) where n is the number of hops 1 2 3 4 5 6 Figure 1.2 An example of string topology and exposed node problem in a wireless mesh network.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 10 23.10.2006 1:43pm & 10 Wireless Mesh Networking when the hop length is less than five hops and beyond five hops, the throughput remains constant albeit at a very low value. Although there exist other factors such as the nature of routing protocol, greediness of the initial nodes and subsequent flow starva- tion of the latter hops, and the behavior of MAC protocols, the single most important factor contributing such a rapid degradation of throughput is the exposed node problem, aggravated by the use of a single-radio system. 1.3.2 Throughput Fairness Another important issue in a single-radio WMN is the high throughput unfairness faced by the nodes in the system. A network is said to be exhibiting high throughput fairness if all nodes get equal throughput under similar situations of source traffic and network load. WMNs show high throughput unfairness among the contending traffic flows especially when CSMA/CA-based MAC protocols are employed for contention 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 MAC protocols, when used in a WMN environment are: (i) information asymmetry depicted in Figure 1.3a, (ii) location-dependent contention depicted in Figure 1.3b, and (iii) half-duplex character of single- channel systems. In Figure 1.3a, only the the receiver of the traffic flow P is exposed to both the sender and the receiver of flow Q, and therefore, the sender of the flow P does not get any information from the channel about ongoing transmissions on other flows. On the other 4 Q Q 4 3 2 5 3 R P 2 1 6 P 1 Coverage of traffic flow P Coverage of traffic flow P Coverage of traffic flow Q Coverage of traffic flow R (a) (b) Figure 1.3 Traffic flows and throughput unfairness in WMNs.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 11 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 11 hand, the channel activity is known to be the sender of flow Q. This information asymmetry causes unfair sharing of the total throughput achieved. 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 has packets ready for transmission, upon detecting an idle channel, it may start transmission by sending request-to-send (RTS) packet to node 2. At this point, if there is an ongoing transmission between nodes 3 and 4, node 2 does not respond to the RTS, leaving node 1 to exponentially back off and retry again. This repeated back-off and several retransmission attempts lead to achieving a low throughput for flow Q. On the other hand, a similar situation can happen to node 3 with a much lower probability and that is proportional to the vulner- able period of the medium access scheme, which in this case is the propagation 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 to throughput unfairness. For example, in Figure 1.3b, flows P and R do not have information about any other flows in the network whereas flow Q has information about both the other flows. Therefore, flow Q has to set its network allocation vector (NAV) and abstain from trans- mitting, whenever it sees a transmission of control packets or data packets 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 the flow Q is in such a position that it experiences much more contention than the rest of the flows 1 and therefore, flow Q receives only 28% of the total throughput compared with 36% throughput share received by both the flows P and R. In fact, the throughput share of flow Q is inversely proportional to the number of neighbor flows contending for 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 by both 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 a collision for both the flows P and R. This perceived collision may reduce the amount of information, about the flows P and R, available at the flow Q and, therefore, leading to further degradation of throughput fairness. In addition to the information asymmetry and the location-dependent contention, the half-duplex property of a single-interface system isYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 12 23.10.2006 1:43pm & 12 Wireless Mesh Networking Q 3 4 3 4 2 2 P P 1 1 Coverage of traffic flow P Flow over a link with bandwidth B Coverage of traffic flow Q Flow over a link with bandwidth B/2 (a) (b) Figure 1.4 Half-duplex radio interfaces in WMNs. another property that causes high throughput unfairness in a single- radio WMN. Due to the half-duplex characteristics, no node can simultaneously receive and transmit. This is illustrated in Figure 1.4a in which a single half-duplex radio with a channel data rate of B bits/s is employed and therefore, only one communication could be permit- ted at any time. Therefore, only flow P could be transmitting while other nodes are waiting. As mentioned earlier, in certain MAC proto- cols such as CSMA/CA-based IEEE 802.11, there exists a strong chance of channel capture where a successful node keeps getting transmis- sion opportunities more often than others. Such channel capturing and subsequent unfairness can be prevented by using multiple chan- nels. In Figure 1.4b, each node uses two radio interfaces with channel data 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 the bandwidth, the throughput fairness increases as found in experimen- tal studies 8. 1.3.3 Reliability and Robustness Another important motivation for using WMNs and especially the MR-WMNs is to improve the reliability and robustness of communica- tion. The partial mesh topology in a WMN provides high reliability and path diversity against node and link failures. MR-WMNs provide the most important ingredient for robustness in communication— diversity. For example, in wireless systems channel errors can be very high compared to wired networks; therefore, graceful degradation ofYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 13 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 13 communication quality during high channel errors is necessary. This is particularly important when the WMN system utilizes unlicensed fre- quency spectrum 9. In order to achieve graceful quality degradation instead of full loss of connectivity, WMNs can employ frequency diver- sity, by using multiple radio interfaces, which is difficult to achieve inasingle-radio WMN system. MR-WMNs can use appropriate radio- switching modules to achieve fault tolerance in communication either by switching the radios, channels, or by using multiple radios simultaneously. 1.3.4 Resource Management Resource management refers to the efficient management of network resources such as energy, bandwidth, interfaces, and storage. For example, the energy resources can be efficiently used in a WMN with limited energy reserve if each node in the system has a new low-power interface in addition to the regular interface. The overall power consumption, even in idle mode, depends very much on the type of interface. Therefore, in an IEEE 802.11-based WMN with limited energy reserve, an additional low-power and low-data rate interface can be used to carry out-of-band signaling information to control the high-power and high-data rate data interface. Bandwidth resources can also be managed better in a multiradio environment. For example, the load balancing across multiple interfaces could help preventing any particular channel getting heavily congested and hence becoming a bottleneck. In addition to balancing the load, bandwidth achieved through each interface can be aggregated to obtain a high effective data rate. In such a bandwidth aggregation mechanism (also known as bandwidth striping), dynamic packet scheduling can be utilized to obtain a better performance. Finally, one important advantage of using a multiradio system in a WMN is the possibility to effect provisioning quality of service through service differentiation. 1.4 DESIGN ISSUES IN WIRELESS MESH NETWORKS There are many issues that need consideration when a WMN is designed for a particular application. These design issues can be broadly classified into architectural issues and protocol issues. The architectural design issues and protocol design issues are described in Section 1.4.1 and Section 1.4.2.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 14 23.10.2006 1:43pm & 14 Wireless Mesh Networking 1.4.1 Network Architectural Design Issues AWMN can be designed in three different network architectures based on the network topology: flat WMN, hierarchical WMN, and hybrid WMN. These categories are briefly discussed below. Flat Wireless Mesh Network In a flat WMN, the network is formed by client machines that act as both hosts and routers. Here, each node is at the same level as that of its peers. The wireless client nodes coordinate among themselves to provide routing, network configuration, service provisioning, and other application provisioning. This architecture is closest to an ad hoc wireless network and it is the simplest case among the three WMN architectures. The primary advantage of this architecture is its simplicity, and its disadvantages include lack of network scalability and high resource constraints. The primary issues in designing a flat WMN are the addressing scheme, routing, and service discovery schemes. In a flat network, the addressing is one of the issues that might become a bottleneck against scalability. Hierarchical Wireless Mesh Network In a hierarchical WMN, the network has multiple tiers or hierarchical levels in which the WMN client nodes form the lowest in the hier- archy. These client nodes can communicate with a WMN backbone network formed by WMN routers. In most cases, the WMN nodes are dedicated nodes that form a WMN backbone network. This means that the backbone nodes may not originate or terminate data traffic like the WMN client nodes. The responsibility to self-organize and maintain the backbone network is provided to the WMN routers, some of which in the backbone network may have external interface to the Internet and such nodes are called gateway nodes. Hybrid Wireless Mesh Network This is a special case of hierarchical WMNs where the WMN utilizes other wireless networks for communication. For example, the use of other infrastructure-based WMNs such as cellular networks, WiLL net- works, WiMAX networks, or satellite networks. Examples of such hybrid WMNs include multihop cellular networks 2, throughput enhanced wireless in local loop (TWiLL) networks 3, and unified cellular ad hoc networks 4. A practical solution for such a hybrid WMN for emergency response applications is the CalMesh platform 5. These hybrid WMNs may use multiple technologies for both WMNYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 15 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 15 backbone and back haul. Since the growth of WMNs depend heavily on how it works with other existing wireless networking solutions, this architecture 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-wise manner starting from the physical layer to the application layer. Some of these protocol design issues are presented below. Physical Layer Design Issues At the physical layer, the main design issue is the choice of an appropriate radio technology. The choice of a radio technology can be based on: (i) technological considerations and (ii) economic con- siderations. The main technological considerations include the spec- tral efficiency, physical layer data rate, and the ability to operate in the presence of interference. For example, the choice of technologies such as code division multiple access (CDMA), ultra wide band (UWB), and multiple input multiple output (MIMO) are more suitable for WMN physical layer than the most popular physical layer technol- ogy, orthogonal frequency division multiplexing (OFDM) used in today’s WMNs. For example, today’s physical layer technology, pri- marily based on OFDM provides a maximum physical layer data rate of 54 Mbps. In a highly dense network with high interference, this capacity may not be sufficient. Therefore, development of new and high 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 the WMN physical layer design. This is emphasized by some of the applications of WMNs such as emergency response and military applications where the spectrum used for communication depends on the unused spectrum in a given locality. In such applications, a software-defined radio with cognitive capabilities would be an ideal choice. In addition to the technological considerations mentioned above, the second most important requirement is economical or social where the simplicity of the physical layer technology will lead to inexpensive devices and hence better social affordability of WMNs. An example of this is evident in the success of today’s IEEE 802.11b- based WMNs where the inexpensive network interface cards contrib- uted to the success of the proliferation of WMNs. Therefore, while choosing the physical layer technology, a network designer should look at the application and user scenario as well.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 16 23.10.2006 1:43pm & 16 Wireless Mesh Networking Medium Access Control Layer The design of MAC layer protocol assumes significance in a WMN because achievable capacity depends heavily on the performance of MAC protocol. In addition to a fully distributed operation, the major issues faced by the popular CSMA/CA-based IEEE 802.11 distributed coordination 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 the network capacity, multiple radios operating in multiple channels are used. Therefore, new MAC protocols are to be designed for operating in multichannel MR-WMN systems. MAC protocols are also to be adapted to operate in different physical layer technologies such as UWB and MIMO physical layers. Another popular research issue for better MAC performance is the use of cross-layer interaction mechan- isms that enable the MAC protocol to make use of information from other layers. In traditional wireless or wired networks, each layer works with its own information making it unable to make the best use of the network-centric properties. In general, the MAC layer protocol design should include methods and solutions to provide better network scalability and throughput capacity. Network Layer Unlike the routing protocols for ad hoc wireless networks, the routing protocols, depending on its network scenario, face different design issues in a WMN. Since WMN is relatively a static network, the routing can make use of table-driven routing approaches such as that used in wired networks or in ad hoc wireless networks 12. The main issues faced 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 adaptabil- ity. The routing metric design plays a crucial role in achieving good performance. The best routing metric may also differ in its perform- ance. For example, in WMNs, the routing metric design has to take the link level signal quality into account for better end-to-end perform- ance. Routing protocols for WMNs, while providing a good end-to-end performance, should also consume minimum bandwidth for setting up paths. In addition, the use of wireless medium demands quick path reconfiguration capability in order to maintain the robustness of the path. Another important aspect is the load-balancing cap- ability that needs to be incorporated with the routing protocol. Finally, a routing protocol for WMN must be adaptable to the networkYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 17 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 17 dynamics. The routing protocol can be classified into either flat routing protocol or hierarchical routing protocol based on the type of network where the routing protocol is applied. Transport Layer At the transport layer, the biggest challenge is the performance of transport protocols over the WMN. Since a WMN has large round-trip time (RTT) variations and these RTT variations are dependent on the number of hops in the path, the end-to-end TCP throughput degrades rapidlywiththroughput.Thepacketloss,collision,networkasymmetry, andlink failures can also contributeto the degradation in transportlayer protocolperformance.ThepopulartransportlayerfortheInternet,TCP, performs very poorly in its original form over a WMN. The transport layer needs to be refined or rewritten for making it more efficient on a WMN. Some of the design issues for a transport layer protocol for WMN are: (i) end-to-end reliability, (ii) throughput, (iii) capability to handle network asymmetry, and (iv) capability to handle network dynamism. Application Layer The most popular application for WMNs is the Internet access service. Essentially, a WMN needs to provide Internet services for residential areas or businesses. In such a situation, though data services make primary service over a WMN, voice services such as voice over Inter- net protocol (VoIP) are also important. Therefore, it is very essential to provide support for both the time-sensitive and the best-effort traffics. In addition to the basic data and voice traffic support, the network provides service discovery mechanisms. Since most of the network services are in fully distributed form, static service discovery mechan- isms may not be effective in a WMN. Another important requirement for the application layer protocol design is to handle the heterogeneity of networks as the data may pass through a variety of networks before being delivered to the end application. System-Level Design Issues The above-mentioned issues are generic to a WMN and these issues are 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 system design, (ii) design for security and trust, (iii) network management systems, and (iv) network survivability issues. Some of the primary challenges faced by a WMN can be alleviated by the use of an MR-WMN and therefore, subsequent sections focus on this.Yan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 18 23.10.2006 1:43pm & 18 Wireless Mesh Networking 1.5 DESIGN ISSUES IN MULTIRADIO WIRELESS MESH NETWORKS The primary advantages of using an MR-WMN are the improved capacity, scalability, reliability, robustness, and architectural flexibility. Notwithstanding the advantages of using a multiradio system for WMNs, there exist many challenges for designing an efficient MR-WMN system. This section discusses the issues to be considered for designing an MR-WMN. The main issues can be classified into archi- tectural design issues, MAC design issues, routing protocol design issues, 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 perform- ance objectives of an MR-WMN when a network is deployed. In general, the network architecture of an MR-WMN is designed on the basis of the type of application or deployment scenario. The major architectural choices to be considered are: (a) topology-based, (b) technology-based, and (c) node-based. Based on the topology, an MR-WMN can be designed either as a flat-topology-based or as a hierarchical-topology-based. The design categories under the technol- ogy-based solution are homogeneous or heterogeneous. While the most popular form of MR-WMN system is the homogeneous type that uses only one type of radio technology such as the popular WLAN technology IEEE 802.11, it is possible to develop an MR-WMN with heterogeneous technologies that utilize a variety of communication technologies. Finally, the node-based design criteria can be classified into either host-based, infrastructure-based, or hybrid MR-WMNs. In the case of host-based MR-WMNs, the network is formed by the host nodes and is same as an ad hoc wireless network with limited or no mobility. On the other hand, in the infrastructure-based MR-WMNs, the WMN is formed by nodes placed on fixed infrastructures or buildings. An example for this architectural type is the rooftop net- works formed by placing wireless mesh relay nodes on the roof of every house for building a residential communication network. Finally, a hybrid MR-WMN has both infrastructure-based backbone and wireless mesh hosts. These hosts communicate over the wireless mesh backbone. This backbone topology can be organized either as a flat topology or as a hierarchical topology as discussed in Section 1.4.1. In some application environments, the hosts are mobile and they also relay traffic on behalf of other hosts in the network. An example of such hybrid MR-WMNs is the vehicular WMNs that communicate overYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 19 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 19 a wireless mesh infrastructure. Therefore, the design of an MR-WMN system 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 main challenges among them are the interchannel interference, interradio interference, channel allocation, and MAC protocol design. The inter- channel interference refers to the interference experienced at a given channel due to the activity in neighbor channels. For example, in IEEE 802.11b, although there are a total of 11 unlicensed channels in North America (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 given geographical location. Therefore, the presence of multiple radios must consider the interchannel interference as the use of a new channel, at a second interface that interferes with the existing channel, will lead to significant 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 and implementation of radio interfaces. This type of interference is experi- enced at a particular radio due to the channel activity at another interface in the same WMN node. Such interferences occur even when both the interfaces use nonoverlapping channels 27. For example, when interfaces A and B on a WMN node use channels 1 and 11, respectively, interradio interference may experience. This interference is primarily due to the design of the hardware compon- ents and the interface itself where usually a number of low-cost filters and associated RF components are used. The physical separation of interfaces may help to avoid this issue to some extent; in certain cases the separation may be difficult, especially in portable nodes. The use of certain low-cost interface cards leads to interference even when they are separated for few feet 27. Another issue of importance to MAC is the channel allocation. This is a network-wide process where the allocation of noninterfering channels would lead to significant throughput and media access performance. The channel allocation should consider the number of channels available and the number of interfaces available. Therefore, techniques such as graph coloring are used for generating channel allocation strategies. Finally, the most important issue is the design of MAC protocols. The availability of multiple interfaces and multiple channels leads to new designs for medium access protocols that can be benefitted in the presence ofYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 20 23.10.2006 1:43pm & 20 Wireless Mesh Networking multiple radios. Examples of such protocols are the multichannel carrier sense multiple access (MCSMA) 24, interleaved carrier sense multiple access (ICSMA) 8, and the two-phase time division multiple access (2P-TDMA) 26. These protocols utilize multiple channels simultaneously and also attempt to solve the media access issue in MR-WMNs. 1.5.3 Routing Protocol Design Issues Another important issue in designing an MR-WMN is the design of routing protocol that depends on the design of the WMN architecture and in some cases it also depends on both the network’s application and the deployment scenario. The routing protocol design can be classified into several categories based on: (a) the routing topology, (b) the use of a routing backbone, and (c) the routing information maintenance 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 is built among the nodes in such a way that the pathfinding responsibil- ity is delegated to higher-level nodes in the hierarchy when the lower level 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 inbuilt hierarchies and each node has equal responsibility to find a path to the destination and to participate in the pathfinding process of other nodes. The chosen path may include any arbitrary node in the net- work without following any particular node hierarchy. Second design category is routing based on routing backbones and is classified into tree-based backbone routing, mesh-based backboneless routing, and hybrid topology routing. Unlike a wireless ad hoc network, a WMN is relatively static or has limited mobility network; therefore, in order to increase the routing efficiency, a routing backbone can be built. An example of this routing approach is the WMN routing performed by the IP routing mechanism over a spanning tree protocol (STP)-based tree backbone 10. In the case of STP, the link layer will form a tree topology among the WMN nodes similar to a wireless distributed system (WDS) 12 and at the network layer, routing is carried out by traditional IP-based routing method. Although this is one of the simplest approach for WMNs, it has several issues such as poor reliability and lack of network scalability. On the other hand, routing protocols designed for and implemented at the network layer may follow a backboneless mesh routing approach. A third approach is to use a network layer backbone-topology, a subset of the nodesYan Zhang / Wireless Mesh Networking AU7399_C001 Final Proof page 21 23.10.2006 1:43pm & Wireless Mesh Networks: Issues and Solutions 21 forming a mesh-like backbone within the WMN, optimized for certain parameters such as throughput, channel quality, or network scalability can be used for aiding a backboneless mesh routing protocol. Such a routing approach that uses a dynamic backbone topology at certain specific segment of the network is called a hybrid topology routing protocol. Finally, routing protocols can be designed on the basis of the routing information maintenance approach. Examples of such routing schemes are proactive or table-driven routing protocols, reactive or on-demand routing protocols, and hybrid routing protocols. In the case of proactive or table-driven routing approach, every node exchanges its routing information periodically and maintains a routing table, which contains routing information to reach every node in the network. Examples of routing protocols that use this design approach are DSDV 13, WRP 14, and STAR 15. On the other hand, in the reactive or on-demand routing approach, a node requests routing information 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 and on-demand routing approaches. An example of such a hybrid routing approach is the zone routing protocol (ZRP) 18, which employs a table-driven routing approach within a zone and on-demand approach beyond the zone. That is, every node uses proactive approach within a k-hop routing zone and employs a reactive routing approach beyond the routing zone. 1.5.4 Routing Metric Design Issues In addition to designing a routing protocol, another important issue is the 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 on which a routing decision is made. Hop count is the simplest routing metric 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 routing protocol and the design of routing metrics should take several factors such as (i) the network architecture, (ii) the network environment, (iii) the extent of network dynamism, and (iv) the basic characteristics of the routing protocol into account, in order to design an efficient routing protocol for WMNs. First, the architectural property of the

Advise: Why You Wasting Money in Costly SEO Tools, Use World's Best Free SEO Tool Ubersuggest.