World Applied Sciences Journal 30 (7): 870-886, 2014 ISSN 1818-4952 IDOSI Publications, 2014 DOI: 10.5829/idosi.wasj.2014.30.07.

1465

A Survey on Routing Algorithms and Routing Metrics for Wireless Mesh Networks
Mohammad Siraj PSATRI, College of Engineering, King Saud University, P.O. Box: 800, Riyadh 11421
Submitted: Nov 30, 2013; Accepted: Feb 10, 2014; Published: Feb 23, 2014

Abstract: Wireless Mesh Network (WMN) is an emerging wireless technology for various futuristic applications like video on demand, healthcare systems, disaster and emergency, modern intelligent transport systems, public security systems, broadband internet for home and campus networking. Routing is an important factor for forwarding a data packet from source node to destination node.Wireless routing is totally different than routing in a wired network. Routing metrics are a crucial part of routing protocols affecting the performance of wireless mesh networks (WMN). When the routing protocols are implemented, the routing metrics are allocated to various paths. Their job is to compute the best routing path. They are assimilated in routing protocols to enhance WMNs efficiency in terms of reliability, latency, throughput, error rate and cost. This paper presents a survey of important routing protocols and routing metrics for WMNs. Key words: Interference Routing protocol Routing metric Wireless Mesh Network

INTRODUCTION IEEE 802.11 Multi Hop Wireless Mesh Network (WMN) [1, 2] is a promising technology, seen as a promising potential for Internet Service Providers (ISPs) and other end-users to deliver a consistent wireless broadband service access and a robust network at a rational cost. They are now being seen as last few miles connectivity. Wireless networks differ from wired networks. They can organize and configure themselves dynamically. i.e., establishment and maintenance of the network is done automatically by the mesh nodes. This automatic establishment and maintenance has numerous benefits for the end-users. Service coverage becomes dependable and network is robust. Deployment and maintenance of WMN is easy and economical. Mesh networks comprises of mesh clients and mesh routers. The end users can use WMN to access the internet by connecting to any of the routers.The client could be a notebook, PDAs, desktop PCs or a mobile phone whereas mesh routers are a part of backbone. They differ from mobile ad-hoc network (MANET), as the problems in MANETs are caused by the absence of infrastructure and nodes mobility [3]. In WMNs connectivity between the mesh clients and mesh routers are through multiple hops.

Generally, the mesh routers are static. Figure 1 shows an architecture of WMN. Routing metricsare very essential for calculation of the best quality path. It does so by capturing an accurately good quality link. Due to the presence of static nodes and common wireless medium in WMNs designing a good routing metric is challenging.To study the impact of a good routing metric design is a challenge for researchers. To understand these challenges the current routing metrics which have been presented for WMN protocols are analyzed and a comparison of their prominent features have been done for their advantages and drawbacks. The remainder of this paper is organized as follows. In section 2 important design requirements of an efficient mesh routing protocol are discussed. In section 3, the review of important routing protocols for WMNs have been discussed. Section 4 discusses the design requirements and characteristics of routing metrics. Section 5 describes a review of important routing metrics for WMNs followed by a conclusion. Requirements of an Efficient Mesh Routing Protocols: An efficient mesh routing protocol should have the following features.

Correspoding Author: Mohammad Siraj, PSATRI, College of Engineering, King Saud University, P.O. Box: 800, Riyadh 11421

870

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 1: Wireless Mesh Architecture [4] Distributed: The protocol should be a distributed, as centralized routing involves higher control in order to have improved reliability. As node mobility is very minimal, therefore, mesh networks prefer the decentralized routing protocol. Adaptable to Topology Changes: The algorithm should adapt to minimal changes in the topology caused by node's mobility. This is a substitute to the uniformly distributed traffic within the network and needs maintenance of the routing paths of all the nodes continuously. Loop-Free: This is a fundamental requirement of any routing protocol to avoid unnecessary wastage of bandwidth. In the mesh network, due to node mobility, loops may be formed during the route establishment. So a routing protocol should eliminate such loops. A routing protocol should be loop-free in order to avoid wastage of bandwidth. Security: If the security doesnt exist, then the routing protocol issusceptible to different types of attacks such as spoofing and redirect messages. To prevent such vulnerabilities, security schemes are required. Cryptographic mechanisms are utilized to solve security issues and prevent such attacks. Scalability: Scalability is the protocol capability to maintain its performance with an increase of users and 871 network size. This requires that control overhead should be minimized and routing protocol should be able to adapt to the network size. In [5, 6] researchers showed that scalability declines with the large number of users. Gupta and Kumar [7] computed per node capacity. The topology of the network decides whether the network would be able to scale or not [8-10]. The network coverage area results in an increase of load. In [11-13], showed how to keep the deterioration to a minimum level. The effect of physical layer on scalability was shown in [14]. It was seen thatdirectional antenna increases performance and scalability [15-16]. Other researchers [17-20] have also studied the effects on scalability on WMN. Quality of Service (QoS): In WMNs, the quality of service (QoS) features requires a great amount of challenge due to resource limitations and node movement. Therefore, it is essential that QoS metrics are incorporated in WMN routing protocols for route discovery and maintenance to support end-to-end QoS. QoS main focus is on end to end delay, bandwidth, Packet delivery ratio and energy and mechanism overheads. Hence, it is mandatory for WMNs to have a proficient routing and QoS mechanism for supporting various applications. The routing protocol should ensure the required level of quality of service and that too should also be in real time to support current traffic. Some of the researchers [21-28] have focused their work to study and improve QoS in WMN.

World Appl. Sci. J., 30 (7): 870-886, 2014

Routing Protocols: The routing protocols can be categorized into proactive routing, reactive routing and hybrid routing protocols. Some of the important protocols in these categories have been discussed below. Proactive Routing Protocols: Proactive routing broadcasts periodic HELLO messages, like traditional routing in the internet, in order to determine the global view of the network topology, which is useful when route establishments are needed. However, established routes, which are cached in each node might never be used. This leads to wastage of network bandwidth, especially in high node density. In addition, in proactive routing protocols, there is a trade-off between the freshness of cached routes and the frequency of message broadcasts. Frequent broadcast messages are useful in order for the packet carrier node to calculate efficient routes to the specified destination. However, this is at the expense of high bandwidth consumption, which grants the channel for broadcast traffic. On the other hand, this type of routing is suitable for real time applications (delay sensitive services) since the route between a pair of sources and destinations is created beforehand. In other words, the source does not need to flood route discovery requests on demand as the route is established in the background. In spite of the low end-to-end latency of packet forwarding, the recovery of unused cached routes wastes massive bandwidth, especially in high mobile environments. Few key examples of proactive routing protocols are Destination-Sequenced Distance Vector Routing (DSDV) [29], Fisheye State Routing (FSR) [30], Optimized Link State Routing (OLSR) [31], Source Tree Adaptive Routing (STAR) protocol [32], HEAT Protocol [33], Wireless Routing Protocol (WRP) [34], Mobile Mesh Routing Protocol (MMRP) [35], Linked Cluster Architecture (LCA) [36], Hierarchical State Routing protocol (HSR) [37], Topology Dissemination based on Reverse-Path Forwarding routing protocol (TBRPF) [38], Direction Forward Routing (DFR) [39] and Distributed Bellman-Ford Routing Protocol (DBF) [40]. In this paper only a few important proactive protocols are discussed. Destination Sequenced Distance Vector Routing Protocol (DSDV): Destination Sequenced Distance Vector Routing Protocol is proactive unicast routing protocol. DSDV is based on traditional Bellman Ford algorithm. In DSDV, every node maintains a routing table. Each entry in the routing table keeps information of all likely node destinations in the network. It also keeps a record of the 872

number of hops for every destination. To avoid loops, DSDV uses sequence numbers. The routing updates are either event driven or time driven. Each node transmits routing table updates and its routing information to its adjacent neighbor nodes periodically. In DSDV, two types of updates are possible. The first one is full dump and an incremental update. In the full dump, the complete routing information is carried and may require the number of Network Protocol Data Unit (NPDU). In contrast, only entries of available destinations with recent changes from the routing table are sent in the incremental update. When the nodes are relatively static, incremental update avoids extra traffic compared to the full dump update. However, the full dump update is more efficient in the network with high speed mobile nodes. In both update mechanisms, the route update packet is sent with a unique sequence number with the routing information. During the selection, the path having the greatest sequence number is selected as the current path. If two paths have the same sequence number, the shortest path is selected. Optimized Link State Routing Protocol (OLSR): Optimized Link State Routing Protocol is built over the link state algorithm. In this protocol, information about the link state of every node is broadcasted to every other node of the network. In OLSR, all nodes keep a track of the information of their two hop neighbors. HELLO messages are used by OLSR to get the information about the links. Multi Point Relays (MPR) is a vital feature of the OLSR protocol to minimize broadcasting. Each node selects a set of MPR among its one hop bidirectional link neighbors to all other nodes those are two hops away. This set can change with timeand is specified by the selecting nodes in their HELLO messages. Whenever a node transmits a message, all of its neighbors receive the message. Only that MPR which receives the message for the first time, transmits the message. Due to this, the overhead due to flooding is minimized. Two types of control messages are used by OLSR. They are HELLO and TOPOLOGY CONTROL (TC). TC messages are periodically sent. The TC messages can be forwarded only by MPR hosts. The major advantage of OLSR over other proactive protocols is that it broadcasts its link state information rather than routing tables and messages can be delivered in any order due to the sequence number. This protocol is good for large and dense networks.

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 2: MRP route establishment message sequence [9] Mesh Networking Routing Protocol (MRP): Mesh Networking Routing Protocol (MRP) is a proactive protocol. In this protocol, to connect to the internet every client chooses a gateway. In the eventuality of the node moving away or breaking down of the gateway node a different gateway is selected. The entire traffic movement to the internet is through the gateway. One of the versions of Mesh Protocol is MRP On-Demand. MRP on-demand (MRP-O) [41] is a purely on-demand protocol. This protocol uses the messages like Route Discovery Message (RDIS),Route Advertisement (RADV), Registration Request (RREG), Route Check Packets (RCHK) and Registration Acknowledgment (RACK). The node which is intended to join the network will send RDIS to its neighboring user nodes to find the way to the closest gateway. Here only the source node one-hop neighbors gets the message. The nodes receiving the RDIS message respond by sending a RADV packet containing the information about their current routes metrics. All the neighboring nodes will send the RADV packets with some random delay to avoid collisions. If a new node joins the network, then all RADV packets will be stored by it. Once all the RADVs have been received, it will choose one or more upstream route to perform routing. Figure 2 showsthe MRP route establishment message sequence. The metrics used for this protocol are hop-count, route stability, minimum delay, maximum bandwidth and minimum packet loss. The metrics used for this protocol are hop-count, route stability, minimum delay, maximum bandwidth and minimum packet loss. 873 Fish-Eye State Routing Protocol (FSR): In [30], the authors developed an efficient link state algorithm that maintains the global knowledge of the network topology at each node and disseminates the local information to the direct neighbor nodes instead of the whole network. In FSR protocol, the updates of link state information vary with the distance. Towards the destination. Figure 3 shows the basic operation of FSR. That is, every node defines a boundary around itself. The inner boundary is formed by the closer nodes and they receive the link state information with the highest frequency, whereas the further nodes broadcast the update with lower frequency. Thus, the FSR protocol exchanges the link state information frequently with the vicinity nodes and with lower frequency for the further nodes. In this way, the nodes can get up-to-date link state information about the nearby neighbor nodes. Apparently, there is a trade-off between the reduction of overhead and the staleness of the link state information, leading to suboptimal route selection. Reactive Routing Protocol: The basic operation of reactive routing protocols [42-44] is route discovery from source node to destination node and works in reverse to the on demand Topology-based routing. This routing solution establishes a route when a node makes a request to transmit packets to another node in the network. At this time, the node re-broadcasts the requested route establishment to find the intended destination. When the destination receives the query (or the en-route nodes know the path to it), it responds to the source for route

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 3: FSR protocol basic operation establishment between source and destination. Whenever the source node wants to transmit data packets towards the destination, it floods the network with route request packets. The destination sends a route reply message, after which the source sends the data packet to the destination. The advantage of this approach is that it does not maintain unused routes and reduces bandwidth overhead in the network. Some of the important examples of reactive protocol are Ad Hoc on Demand Distance Vector (AODV) protocol, Temporarily Ordered Routing Algorithm (TORA) protocol [45], Dynamic Source Routing (DSR) protocol [46], ABR (Associativity Based Routing) protocol [47], Link Quality Source Routing Algorithm (LQSR) protocol [48], Dynamic MANET on Demand (DYMO) [49], Lightweight Mobile Routing protocol (LMR) [50], Load-Balancing Curveball Routing (LBCR) [51], Scalable Location Update-Based Routing Protocol (ScrRR ) [52] and Interference-Aware Load-Balancing Routing (IALBR) [53]. Dynamic Source Routing Algorithm (DSR): Dynamic Source Routing Algorithm (DSR) is the source based routing protocol where the source records the sequence of intermediate nodes in a data packet which is sent to the destination. The basic operation of the protocol consists of two phases, namely the path discovery and path maintenance process. In DSR, a path discovery phase is begun when the source node without a valid path intends to transmit a data packet to the destination node. DSR applies the source routing strategy to broadcast a route request message. This includes source id, destination id, a route record with an empty list of addresses of all intermediate nodes and a unique request id towards the destination node. On receiving route request, an intermediate node caches the route record. Figure 4 shows the construction of route cache during node discovery. A route reply message may be replied if the destination 874

node is arrived. The destination node cache stores the route record. Then it uses the cached path in the route record for the propagation of route reply back to the source node as shown in Figure 5. Otherwise, if this node is not in the route cache of route request, it appends its address to the route record and broadcasts the route request messages. To avoid the overhead, DSR optionally defines the unique request id for each message in the route discovery mechanism. In addition, these messages are forwarded hop-by-hop. Unlike other on-demand driven algorithms, there are no proactive periodic probes for neighbor detection, or link status detection in DSR. Therefore, DSR operates truly on demand to minimize the routing overhead. The route maintenance mechanism is initiated when a node cannot deliver packet to its next-hop node as shown in Figure 6. This node then generates route error messages towards the source node to find the most viable route as shown in Figure 7. Hence, the broken link is removed from the route cache of the source node. Ad-Hoc On-Demand Distance Vector Routing Algorithm (AODV): AODV is a reactive unicast routing protocol. AODV protocol doesnt deploy flooding. AODV does not store routing information of all the nodes, instead it just keeps information about the nodes falling on the active route. In AODV when a source node has data packets and intends to communicate with another node, it initiates the route discovery process in the network. As the source has no suitable route for the destination, it broadcasts the Route request (RREQ) message as shown in Figure 8. RREQ packet contains the address of the source node, address of the destination node and broadcast id. Broadcast id is an identifier that contains the most-current sequence number of the source and the destination node. Each RREQ begins with the least Time to Live (TTL) value. The TTL value increments by 1 if the destination node is not found.HELLO messages, are used to notify adjacent neighbor nodes. Routing tables keep records for a specific period. A cache is maintained by each node. The cache keeps the entries of the received RREQs. The RREQ with the greatest sequence numbers is accepted and others are rejected. The cache also saves the return path. RREP message is generated and is transmitted back to the source node provided the sequence number of the destination node is equal to or larger. This can be seen from Figure 9.

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 4: Constructing route cache during node discovery

Fig. 5: Route Reply

Fig. 6: Route Error between Node 5 and Node 8

Fig. 7: Route Error sent from N5 to N1 Link Quality Source Routing Algorithm (LQSR): Link Quality Source Routing Algorithm was designed by Microsoft Research Group for their Mesh Connectivity Layer (MCL). Through LQSR, the computers are connected to form a mesh network using WiMax or Wi-Fi. The LQSR protocol is built on Dynamic Source Routing (DSR). 875 LQSR allocates comparative weights to the links between the wireless meh nodes after the nodes have been detected. Besides allocating weights, for every probable link the bandwidth, loss and the channel are calculated. All the nodes are conveyed this information. Depending on this information, LQSR uses Weighted Cumulative Expected Transmission Time (WCETT)

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 8: Broadcasting RREQ Packets

Fig. 9: Route of the RREP to the Source routing metric to find the best routing path from source to the destination node. The path between two nodes may be one hop or more depending whether intermediate nodes exist or not between the source and the destination node. The route is amended accordingly if the optimal path between source and destination is modified. This amendment is done without interruption of the link among the nodes. The LQSR protocol does not require very little administration like DSR protocol to function automatically. In LQSR, after a node gets a RREQ packet, a link quality metric is appended for the link over which the packet is received. When the source node gets a RREP packet, it contains information about link quality and node. To get the information about link state, adjacent nodes are broadcasted HELLO messages by LQSR. These messages are utilized to find out the quality of the link over which the message was got. Temporally-Ordered Routing Algorithm- (TORA): Temporally Ordered Routing Algorithm (TORA) is an on demand routing protocol. It is a distributed routing protocol. It is designed to reduce the communication overhead involved in adjusting to the changes which occur whenever there is a change in network topology. TORA's control messages are generally restricted to a very minor group of nodes. It ensures loop free routes. 876 Generally, TORA offers several routes from source to destination. TORA functions can be broadly classified into three types, which are a) 1) route creation, 2) maintaining route and 3) removal of route. It uses height metric during route creation and maintenance to form a Directed Acrylic Graph. (DAG) fixed at destination. Links are allocated on the basis of relative height of the neighboring nodes. Timing is very crucial in TORA as the height metric relies on link failure logical time. When TORA wishes to delete the routes, it broadcasts a clear (CLR) packet flooding the entire network to delete the invalidated paths. The messages move in a downward flow that is a from a higher height node to a lower height node. Discovery and updating of routes is through Query (QRY) and Update (UPD) packets. The QRY packet circulates over the network until it finds a node which is either the destination node or has route information. A UPD packet containing the nodes height is then broadcasted. On getting the UPD packet each node sets its height larger than the height mentioned in the UPD message. After setting its height higher, this node broadcasts its own UPD packet resulting in many routes. Hybrid Routing Protocol: Hybrid routing protocols combines the advantages of both reactive and proactive routing protocols for performance and scalability.

World Appl. Sci. J., 30 (7): 870-886, 2014

It provides a mechanism such that it implements proactive routing for the nearby and frequently used routes. The reactive routing technique is used for far away nodes and seldom used for data relay. They minimize proactive routing protocol control overhead and reduces the delay in reactive routing protocol during the route discovery process. Examples of hybrid routing protocols are HazySighted Link State Routing (HSLS) protocol [54], Zone Routing Protocol (ZRP) [55], Hybrid Ad-Hoc Routing Protocol (HARP), Hybrid Routing Protocol for Large Scale Ad-Hoc Networks with Mobile Backbones (HRPLS) [56] and Zone-based Hierarchical Link State routing (ZHLS) [57]. In this section only a few important hybrid protocols are discussed. Hazy-Sighted Link State Routing Algorithm (HSLS): Hazy-Sighted Link State Routing Protocol (HSLS) is one of the most efficient hybrid routing protocol for mesh networks. Cowing Foundation is involved in the development of HSLS. Researchers at BBN Technologies created the HSLS routing protocol. The network is not flooded by HSLS. HSLS features are built on reactive, proactive and suboptimal routing methodology. By restricting link state updates in time and space domain, the updates are combined and sent once thereby saving transmission capacity. In HSLS, two proactive algorithms are integrated. The algorithms are Near-Sighted Link-State Routing (NSLSR) and Discretized Link-State Routing (DLSR). NSLSR restricts the number of hops for the communication of routing information. DLSR restricts the number of times a routing information may be communicated. The reactive routing is needed due to a failed attempt in the usage of a nearby link. This causes the expiry of the next timer, maybe to recover the information in finding a substitute route. The reactive routing takes place when there is a failure to use a neighboring link which causes the expiry of the next timer. It then tries to find another route. The main aim is to measure the global network wastage. It includes transmitting route updates of inefficient transmission routes. Researchers used arithmetical optimization technique to compute how much time link state updates take to transmit and what is the coverage of nodes link state updating. Whenever there is a broken connection, a local routing cache update is required. This is the reactive portion of the algorithm. HSLS provides good scalability properties and creates good routes in real time. The data transfer and routing information are distributed, so it provides good reliability and performance. 877

Zone Routing Protocol (ZRP): ZRP falls in the category of a hybrid protocol having reactive and proactive routing characteristics. It creates routing zones based on the neighbor nodes hopping distance. There are two zones created by it. One is the inside zone also called Intra Zone where the packets are sent to the peripheral nodes from the originating node and the other zone which is outside zone also called Inter Zone from where the packet is delivered to the sink node. This is shown in Figure 10 below. The zones are formed from the origin node at two hops away. ZRP is created by two routing protocols, which are Intra-Zone Routing Protocol (IARP) [57], a proactive routing protocol used for intra zone and a reactive routing protocol called Inter-Zone Routing Protocol (IERP) [57], used for Inter Zone, respectively. The path from origin to sink node is made by IARP. Generally, all the current proactive routing algorithms can be used as the IARP for ZRP. For the inter zone paths, it uses IERP to find the route. The originating node generates a route request to peripheral nodes. Peripheral nodes check their zone for the sink node. If the destination node is not a part of this zone, the peripheral node appends its own address to the route request packet and forwards the packet to its peripheral nodes. If the sink node is a member of its zone, it generates a route reply packet on the reverse path back to the source. The originating node saves this path and sends the route reply to the sink nodes. Design Requirements and Characteristics of Routing Metrics: There are four key requirements of routing metrics for ensuring a good performance. First of all, there should be route stability to guarantee that the network is stable. It means that the routing metrics should not change the route frequently. Secondly, WMN characteristics can be captured by the metric for ensuring good performance is on the paths having least weights. Thirdly, for computing, the routing metrics should have proficient algorithms with polynomial complexity to compute paths having the least weight. Lastly, the routing metrics should precisely capture quality links and help in calculating efficient paths while ensuring that routing protocols do not create forwarding loops. For creating a routing metric some important components required are number of hops, channel diversity and link quality. In a WMN, a good routing metric should address the following criteria. Intra-Flow Interference: Intra-flow interference takes place due to the interference of adjacent nodes on the same routing path. They compete against each other for

World Appl. Sci. J., 30 (7): 870-886, 2014

Inter Zone Intra Zone

P
Peripheral Node Originating Node

S
Sink Node

Fig. 10: An example of ZRP Protocol

Fig. 11: Intra-flow interference [58] utilizing the channel. This intra-flow interference causes throughput to degrade severely due to the consumption of the flow bandwidth across every node on the same routing path. The Hop-Count of the flow increases with an increase in the end to end delay. This causes congestion. For example, in Figure 11, it is shown that the path 1 2 3 causes intra-flow interference because of the reuse of channel 1 on the flow 1 2 and from 2 3. So the path 1 4 3 does not have intra-flow interference as two different channels are assigned between 1 4 and 4 3. It can be said, that 1 4 3 is a better path in comparison to 1 2 3. So a good intra-flow aware interference metric should assign 1 4 3 a lower weight than 1 2 3. In other words, a good routing metric reduces inter-flow interference by selection of different channels for neighboring nodes coming on the same path. Inter-Flow Interference: Inter-flow interference occurs due to the other flows operational on one particular channel and nodes are contending for that channel. This is caused by the multiple flows between different routing paths as shown in Figure 12. This consumes the bandwidth of the nodes which are on this route. Furthermore, it competes for bandwidth occupation with the neighboring nodes. In comparison to intra-flow interference, inter-flow interference is harder to control due to the involvement of multiple flows and routes. 878 Figure 12 shows inter-flow interference because of the two paths namely 1 4 3 and 5 6 7. A good inter-flow aware metric should assign a low weight 1 2 3 then to 1 4 3 as path 1 2 3 has less inter-flow interference. Load Balancing: It is defined as the capability of a routing metric to balance the traffic load so that overloading of gateways in wireless mesh networks can be avoided and the network resources are used fairly. WMNs are different than other wired and wireless networks. The environment of the shared channel, presence of stationary mesh node, traffic routing pattern from the user to gateway node and the usage of multi radios distinguishes them from other networks. This unique characteristics and unbalanced load in WMN causes load balancing problem. Load balancing problem becomes an important issue in WMN as the volume of the network traffic is very high. Most of the traffic in WMNs is directed towards the Internet gateway. This may result in an increase in traffic load on some path which are connected to the internet gateway. Due to the unbalanced load, channel overloading, gateway overloading and center loading may occur. Channel overloading occurs when some of the channels are overloaded as compared to other channels. Center loading occurs when the nodes at the centre of the network topology are used more during the traffic flow as they fall on the shortest path. The traffic is oriented

World Appl. Sci. J., 30 (7): 870-886, 2014

Fig. 12: Inter-Flow interference [59] towards the gateways as the traffic is routed through them. Due to the heavy traffic volume, congestion may occur. The focusing of traffic on gateway causes a drop in the packets due to the overflow in the buffer. This dropping of packets is unfavorable for the network to be efficient. This is called gateway overloading. Thus, there is a critical need to avoid congestion due to the unbalanced load. This can be done by load balancing at different gateways and on the paths leading to the gateway. Congested links should be avoided by segregating congested links from the new paths to have load balancing. Load balancing and interference are related. So to solve the problem of unbalanced load, interference problem should also be examined. Isotonicity: It is defined as the characteristic of the routing metric by which it ensures that the order relation between the weights of any two paths is conserved if both are preceded by a third path which is common to the two paths. For example, if W (p) is the weight which is the function along the path p and W (q) is the weight which is the function of path q, isotonicity can be defined as: A routing metric is isotonic if the quadruplet (A, , W, ) is isotonic and if W (p) W (q) implies W(r p) W(r q) as shown in Figure 2.10. Isotonicity is a very crucial requirement of a routing metric for the efficiency of routing protocol. Besides these characteristics, it is essential for a metric to find routes with the maximum throughput constantly. They should have a low packet delivery ratio during the transmission of the packet. The next section will discuss the current routing metrics deployed by WMN routing protocol for their strengths and drawbacks. 879 Routing Metrics: It is a recognized fact that routing metrics are a crucial part affecting the performance of wireless mesh networks (WMN). When the routing protocols are implemented, the routing metrics are allocated to various paths. Their job is to compute the best routing path. They are assimilated in routing protocols to enhance WMNs efficiency in terms of reliability, latency, throughput, error rate and cost. Routing metric is very essential for calculation of the best quality path. It does so by capturing an accurately good quality link. To study the impact of a good routing metric design, is a challenge for researchers. To understand these challenges the current routing metrics which have been presented for WMN protocols are analyzed and a comparison of their prominent features have been done for their advantages and drawbacks. Hop-Count Metric: The most common metric used in the multi-hop routing protocols is Hop-Count metric. It is used in protocols like DSDV, DSR and AODV. Hop-Count metric finds route having the minimum number of hops. Hop-Count can outclass other metrics, which are dependent on load in high agility situations. This shows its agility. The metric is very stable and also has the isotonicity characteristic. As a result, least weight paths can be discovered proficiently. The weakness of this metric is that it does not address interference, channel diversity, varying load of the link and capacity of the load. These are the factors that are experienced by the links. It treats all the links identical. It finds path having poor throughput and high packet loss ratio. This is because the links which are slower, take a lot of time to transmit packets.

World Appl. Sci. J., 30 (7): 870-886, 2014

p r q
Fig. 13: Isotonicity W(p) W(q)

q
W(r p) W(r q)

Expected Transmission Count (ETX) Metric: The ETX metric [60] was presented to address the problems faced by Hop-Count metric. The ETX is the number of transmissions required to successfully deliver a packet over a wireless link at the MAC layer. The ETX of a path is the summation of ETX of every link over the path. In mathematical terms, it can be written: P= 1-(1-P f) (1-P r) (5.1)

where, P = Probability of unsuccessful transmission of packet in a link from node a to node b. Pf = Probability of path loss in forward direction. Pr = Probability of path loss in reverse direction. The expected number of transmissions to successfully deliver a packet in one hop can be represented by.
ETX = 1 kpk (1 p)k = k =1 1 1 p

computation of loop free routes and least weight as it is isotonic. The weakness of this metric is that it is not agile and is meant for single channel only. It does not consider links interference and only captures a link loss ratio thereby compromising link quality affects the link data rate transmission. It does not calculate traffic loss rate precisely as it doesnt capture variation in rates of transmission. It leads to an unbalanced load of the network, because it does not consider varying link load and routes through heavily loaded nodes. ETX doesnt minimize intra-flow interference as it is not able to differentiate among diverse channel routes and same channel routes. Expected Transmission Time (ETT) Metric: ETT metric [61] was an enhancement over ETX as it took into account the bandwidth of different links. ETT is the time taken to communicate a packet successfully to the MAC layer.
ETT = ETX * S B

(5.2)

(5.4)

The ETX metric for a single link is measured in terms of forward and reverse delivery ratio.
ETX = 1 M f * Mr

where, S = Packet size (average) and B = Bandwidth of the current link. ETT metric is got by summation of all the ETT values of the different links on the path. The advantage of ETT is, that it is isotonic and increases the performance of the whole network by increasing the path throughput by determining link capability. It contains all the same drawbacks of ETX. ETT is unable to avoid the routing through the nodes and links which are severely loaded. It does not reduce inter- flow and intra-flow interference as it has not been designed for multi radio networks. Weighted Cumulative Expected Transmission Time (WCETT) Metric: WCETT [61] was presented toenhance the ETT metric in the multi radio mesh networks by taking into account the diversity of the channels. 880

(5.3)

where, Mf = Forward delivery ratio is (1-P f) Mr = Reverse delivery ratio is (1-Pr) ETX measures the packet loss rate. Every one second, probe packets are sent to all neighboring nodes. On getting the probe packet, the neighboring node sums the number of packets received. Based on this information, every ten seconds packet loss rate is computed. It finds paths, which have a high throughput with the minimum hops because a long path will have less throughput because of intra-flow interference. It indirectly handles inter-flow interference. It allows proficient

World Appl. Sci. J., 30 (7): 870-886, 2014

= MIC ( p ) The WCETT metric of a path p is defined as follows: 1 N * min( ETT )

IRU L + CSCi
i p i p

(5.7)

WCETT p = (1 ) * ETT +

* MaxX j

(5.5)

where, Xj = Summation of links ETT values which are on channel j in a system having orthogonal channels.
= Xj

hops on channel j

ETTi 1 j k

(5.6)

N = Total nodes and min (ETT) = Networks least ETT value It consists of components of MIC, IRU (Interference Aware Resource Usage) and CSC (Channel Switching Cost). IRUL = ETTL * NL. CSCi = w1 if Chi1 = CHi CSCi = w2 if Chi1 Chi, 0

is a tunable parameter between 0 1 which controls the preferences over the path length versus channel diversity. WCETT is a weighted average of two components. The first term is usually the summation of the individual link ETTs while the second term is the summation of the ETTs of every link of a given channel. This adds channel diversity to the routing metric causing low intra-flow interference. Using WCETT, multi radio wireless mesh network's performance is enhanced in comparison to ETX, ETT and Hop-Count metrics. WCETT metric is not isotonic and due to this, it cant be used with link state routing protocols. Secondly, the inter-flow interference effects are not explicitly taken into account by WCETT. As a result of this, sometimes paths are created, which have high levels of interference. WCETT is very effective in selecting paths having channel diversity as it takes intra-flow interference into consideration. It retains the gains of ETT metric barring isotonicity.Multi-radio wireless mesh network performance is enhanced after using WCETT in comparison to Hop-Count, ETX and ETT metrics. It maintains a balance between delay and throughput. WCETT does not consider the locality information of the links instead, it considers those links which are operational on that channel. It assumes that there is an interference by the links which are operational on the same channel which may lead to congested paths. It is not isotonic as it can be seen by the presence of the second term. As it does not have isotonicity, it is not easy to use for link state routing protocols. This metric has the same limitations as ETT/ETX metric as it does not estimate actual link presence. Also it does not take into account inter-flow interference effect. As a result, routes with high interference may be established. Metric of Interference and Channel Switching (MIC): MIC metric [62-63] is an isotonic metric and designed to consider inter and intra-flow interference effects besides providing load balancing. For a path p 881

w1 < w2

NL is the group of neighbors nodes, which interferes with communications on link I. CH i denotes channel allocated for ith node communication and i-1 denotes the earlier hop of i th node on the path p. MIC addresses both types of interference. Breaking into imaginary nodes through a least weight algorithm like Dijkstra's algorithm [64], it can be made isotonic. This metric is based upon the assumption that whatever links are positioned in the interference region for a specific link adds an equal amount of interference. It calculates total interference on a particular link through the placement of interference creating nodes irrespective of participating in transmission concurrently or not. Intra-flow interference is captured in two successive links only by the CSC which is the second component. Load Aware Expected Transmission Time (LAETT Metric): There are two main aims of LAETT [65]. The aim is to create a path for fulfilling the flow bandwidth demand and to keep a space for the future needs. It is a combination of load estimation and features of wireless access. It comprises of an implementation of ETT metric.
ETT jk = ETX jk * S B jk

(5.8)

ETXjk = Expected transmission count on the link (j, k) S = Size of the packet Bjk = Bit rate
B jk =
jk jk

Bj jk

Where Bj is the jth node communication rate

= Link Quality Factor = 1 for a very good link

Remaining capacity (RCi) for every node is.

RC = j Bj

g jl
k =1

jl

World Appl. Sci. J., 30 (7): 870-886, 2014

fjl is the rate of transmission of current flow Nj that travels across jth node. The flow cost on the leftover bandwidth is weighted by factor jl. A superior communication results in a lower usage of bandwidth in comparison to inferior communication. LAETTjk is defined by:
LAETT jk = ETX jk * S RC j + RCk 2 jk

MTI metric can be defined by the following equation. MTIj (Q) = ETTjk (Q) * AILjk (Q), Nl (Q) MTIj (Q) = ETTij (Q), Nl (Q) = 0 AILjk 0

(5.9)

= Neighbors average load which may interfere when communicating amongst nodes j and k using channel Q.
jk = ,N L (Q) NL N L (Q) IL (Q) N j (Q) N k (Q)

The equation 5.9 consists of two parts. The later part captures the leftover bandwidth present on either sides of Iljk (Q) = Neighbors interfering load the node. If there are two paths having identical aggregate NL(Q) = Interfering nodes set of neighbors j and k. ETX weight, LAETT gives importance to the path having is applied to MTI metric for the greatest leftover bandwidth. LAETT advantage is that Scaling factor it is isotonic and load aware. It utilizes shortest weighted balancing the difference in magnitude of two components routing path for balancing the network load. It also (MTI and CSC). can be represented as: captures traffic load and quality of links. The weakness of 1 this metric is that it does not consider intra-flow min( ETT ) * min( AIL), Nl (Q) 0 = interference and is not considered as it does not consider 1 ETT ), N l (Q ) 0 = min( = inter-flow interference. Exclusive Expected Transmission Time (EETT): EETT [66] is an innovative routing metric which is interference aware. It finds multichannel routes having minimum interference to have a high throughput. Multi-channel paths are given better valuation by it. The kth link EETT is given by the following equation:
EETTk = linki IS (k )

AIL= jk (Q)

Min (AIL) = load average and Min (ETT) = least ETT The estimation of interfering nodes' load is a key issue in the deployment. ILA takes into account disadvantages of prevailing metrics like WCETT, ETX, ETT and Hop-Count. It successfully discovers lower congested and a low level interference path which has high throughput and a low packet loss ratio. Inter-flow interference is calculated by ILA. The drawback of this metric is that in two successive links only, the second component CSC can capture intra-flow interference. Interference Aware Routing Metric (iAWARE): iAWARE metric [68] can be said as the first metric which considers both inter-flow and intra-flow interference in WMN. The iAWARE metric can be illustrated as.
n (1 ) * iAware( p) = iAWAREi +

ETTi

(5.10)

The paths weight is the addition of EETTs of complete links falling on that route. EEET has all the benefits of ETT as it has been built over ETT. It is also isotonic. It efficiently takes into account intra-flow interference directly and considers inter-flow interference indirectly. The drawback of this is that it does not consider load variation. Interference Load Aware (ILA): ILAmetric [67] consists of two components: Channel Switching Cost (CSC) and Metric of channel interference (MTI). CSC component can be defined by the following equation. CSCj = w1 if Chj1 = Chj CSCj = w2 if Chj1 Chj, 0 w1 < w2

i =1

* max1 k l X k

(5.11)

Xk is identical to WCETT. For a link k,the iAWARE value is.

iAWAREk = ETTk IRk

CH (j) represents channel assigned for node the transmission and j-1 represents the previous hop of the node j along path p. 882

Irk =Interference ratio among two nodes a and b for a link k.

World Appl. Sci. J., 30 (7): 870-886, 2014

Table 1: Comparison of the existing metrics. Routing Metrics Inter HOP ETX ETT WCETT MIC ILA iAWARE EETT LAETT NO NO NO NO YES YES YES YES NO Intra NO NO NO YES NO YES YES YES NO Link Loss Ratio NO YES YES YES YES YES YES YES YES Over Heads Load Balancing Multi Channel NO NO NO NO YES YES NO NO NO NO NO NO NO NO NO NO NO YES NO NO NO YES YES YES YES YES NO Isoto Nocity YES YES YES NO YES NO NO YES YES Link Data Rate NO NO YES YES YES YES YES YES YES Multi /Single Radio SINGLE SINGLE SINGLE MULTI MULTI MULTI MULTI MULTI MULTI

IRk = min( IRk ( a), IRk (b) SINRl (a ) IR j ( a ) = SNRl (a )

CONCLUSION This paper presents a survey of the routing protocols and routing metrics proposed for wireless mesh networks. Routing algorithms can also be categorized on the basis of geographical, hierarchical and multi-path routing butin this survey only the broadly classified routing protocols for wireless mesh networks have been considered. Also an attempt to present a comprehensive review of popular routing metrics has been done. ACKNOWLEDGEMENT The researcher wish to thank the Deanship of Scientific Research, College of Engineering, King Saud University for supporting this research. REFERENCES 1. Akyildiz, I.F., X. Wang and W. Wang, 2005. Wireless mesh networks: a survey. Journal of Computer Networks Journal, 47(4): 445-487. Bruno, R., M. Conti and E. Gregori, 2005. Mesh networks: commodity multi-hop ad hoc networks, IEEE Communications Magazine, 43(3): 123-131. Chlamtac, I., M. Conti and J. Fr, 2003. Mobile ad hoc networking: imperatives and challenges. Ad Hoc Networks Journal, Elsevier, 1(1): 13-64. Siraj, M. and K.A. Bakar, 2011. Link establishment and performance evaluation in IEEE 802.16 wireless mesh networks. Int. J. Phys., 6(13): 3189-3197. Fong, B., N. Ansari, A.C.M. Fong, G.Y. Hong and P.B. Rapajic, 2004. On the scalability of fixed broadband wireless access network deployment. Proceedings of IEEE Radio Communications, 42(9): S12-S18. Held, G., 2005. Wireless mesh networks. Auerbach Publications, Taylor & Francis Group.

SNRi (a) = for link l node as signal to noise ratio iAWARE can capture varying transmission rates, intra and inter-flow interference and varying link loss ratio effects. This metric preserves all the features of WCETT barring the mechanism of countering inter-flow interference calculation. Average interference produced from neighboring nodes is directly measured. SINR implementation is a big step forward for minimizing inter-flow interference routing in comparison with metrics ETX and WCETT. The weakness of iAware is that it does not possess isotonicity. If a link has higher IRj in comparison to ETTj, iAWARE value is lower as a result a lower ETT path but having greater interference is chosen. The biggest weakness of iAWARE is that it allocates additional weightage to ETT in comparison to interfering links. Besides the routing metrics which have been discussed, other researchers proposed adaptive load aware routing metric (ALARM) [69] and a location aware routing metric (ALARM) [70] considering inter and intra-flow interferences and load balancing on top of being isotonic. A round trip time metric (RTT) was presented which was load dependent but resulted in unstable routes [71]. Medium Time Metric (MTM) [72] considers the overhead in Media Access Control (MAC) layer. An Adjusted Expected Transfer Delay (AETD) metric was proposed, which considers delay and jitter of paths to make a decision, but it falters when there is high interference in the channels [73]. Some other routing metrics presented are Interference-Aware Routing Metric (IAR) [74],Success Probability Product (SPP) [75], Expected Multicast Transmissions (EMT) [76] and Link Aware Metrics [77-78]. The strengths and weakness of the metrics discussed are summarized in Table 1. 883

2.

3.

4.

5.

6.

World Appl. Sci. J., 30 (7): 870-886, 2014

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Gupta, P. and P.R. Kumar, 2000. The capacity of wireless networks. IEEE Transactions on Information Theory, 46(2): 388-404. Jovic, A., P. Vishwanath and S. Kulkarni, 2004. Upper bounds to transport capacity of wireless networks. Proceedings of IEEE Transactions of Information Theory, 50(11): 2555-2565. Jun, J. and M. Sichitiu, 2003. The nominal capacity of wireless mesh networks. IEEE Wireless Communications Magazine, 10(5): 8-14. Li, J., C. Blake, D.S.J. Couto, H.I. Lee and R. Morris, 2001. Capacity of ad hoc wireless networks. In Proceedings of 7th ACM International Conference on Mobile Computing and Networking. Lam, R.K., D.M. Chiu and J.C.S. Lui, 2007. On the access pricing and network scaling issues of wireless mesh networks. IEEE Transactions on Computers, 56(11): 1456-1469. Leveque, O. and E. Telatar, 2005. Information theoretic upper bounds on the capacity of ad hoc networks. Proceedings of IEEE Transactions of Information Theory, 51(3): 858-865. Jelenkovic, P.R., P. Momcilovic and M.S. Squillante, 2007. Scalability of wireless networks. Proceedings of IEEE/ACM Transactions on Networking, 15(2): 295-308. Huang, L.F. and T.H. Lai, 2002. On the scalability of IEEE 802.11 ad hoc networks. Proceedings of the Third ACM International Symposium on Mobile Ad Hoc Networking and Computing (MobiHoc02), 173-182. Peraki, C. and S.D. Servetto, 2003. On the maximum stable throughput problem in random networks with directional antennas. Proceedings of the 4th ACM international symposium on Mobile Ad Hoc Networking & Computing, pp: 76-87. Yi, S., Y. Pei and S. Kalyanaraman, 2003. On the capacity improvement of ad hoc wireless networks using directional antennas. Proceedings of the 4th ACM international symposium on Mobile Ad Hoc Networking & Computing, pp: 108-116. Jiang, W., Z. Zhang and X. Zhong, 2007. High throughput routing in large-scale multi-radio wireless mesh networks. Proceedings of the ECNC, pp: 3598- 3602. Arpacioglu, O. and Z.L. Haas, 2004. On the scalability and capacity of wireless networks with omnidirectional antennas. Proceedings of the IPSN04, Berkeley, California, USA, pp: 169-171. 884

19. Chen, Y., G. Zhu, et al., 2008. On the Capacity and Scalability of Wireless Mesh Networks, Wireless Communications, Networking and Mobile Computing, pp: 1-5. 20. Yonghui Chen, 2011. On the Capacity and Scalability of Wireless Mesh Networks, Wireless Mesh Networks, In Tech, Available from: http://www.intechopen.com/books/ wireless-meshnetworks/on-the-capacity-and-scalability-of-wirelessmeshnetworks 21. Pourfakhar, E. and A.M. Rahmani, 2010. A hybrid QoS multicast framework-based protocol for wireless mesh networks, Computer Communications, 33(17): 2079-2092. 22. Xue, Q. and A. Ganz, 2002. QoS Routing for Mesh-Based Wireless LANs, International Journal of Wireless Information Networks, 9(3): 179-190. 23. Aoun, B., R. Boutaba, Y. Iraqi and G. Kenward, 2006. Gateway Placement Optimization in Wireless Mesh Networks with QoS Constraints, IEEE Journal of Selected Areas in Communications, 24(11): 2127-2136. 24. Drabu, Y. and H. Peyravi, 2008. Gateway Placement with QoS Constraints in Wireless Mesh Networks in Proc. of Seventh International Conference on Networking, pp: 46-51. 25. Junhai, L., Y. Danxia, X. Liu and F. Mingyu, 2009. A Survey of Multicast Routing Protocols for Mobile Ad-Hoc Networks, IEEE Communications Surveys and Tutorials, 11(1): 78-91. 26. Sun, B. and L. Li, 2006. QoS-aware multicast routing protocol for Ad hoc networks, Systems Engineering and Electronics, 17(2): 417-422. 27. Tang, J., G. Xue and W. Zhang, 2005. Interferenceaware topology control and QoS routing in multichannel wireless mesh networks, MobiHoc. 28. Yong, D., K. Pongaliur and X. Li, 2009. Hybrid multi-channel multi-radio wireless mesh networks, IWQoS. 17th International Workshop. 29. Perkins, C. and Bhagwat, 1994. Highly Dynamic Destination-Sequence Distance Vector, Routing (DSDV) for Mobile Computers, in Proc. of ACM SIGCOMM Computer Communication Review, 234244. 30. Pei, G., M. Gerla and T. Chen, 2000. Fisheye state routing in mobile ad hoc networks. In Proceedings of the 2000 International ICDCS Workshop on Wireless Networks and Mobile Computing. 10-10 April. Taipei: Citeseer, D71-D78.

World Appl. Sci. J., 30 (7): 870-886, 2014

31. Jacquet, P., P. Muhlethaler, A. Qayyum, A. Laoutit, L. Viennot and T. Clausen, 2003. Optimized Link State Routing Protocol (OLSR), IETF RFC 3626, http://www.olsr.net/,http://www.olsr.org/. J.J. Garcia and M. Spohn, 1999. Source-tree routing in wireless networks, in ICNP, pp: 273-282. 32. Baumann, R., S. Heimlicher, V. Lenders and M. May, 2007. HEAT: Scalable routing in wireless mesh networks using temperature fields, in IEEE International Symposium on World of Wireless, Mobile and Multimedia Networks. 33. Murthy S. and J.J. Garcia-Luna-Aveces, 1996. An Efficient Routing Protocol for Wireless Networks, ACM/Baltzer Journal on Mobile Networks and Applications, Special Issue on Routing in Mobile Communication Networks, 1(2): 183-197. 34. Grace, K., 2000. Mobile Mesh Routing Protocol (MMRP), http://www.mitre.org/ work/tech_transfer/ mobile mesh/. 35. Gerla M. and J.T. Tsai, 1995. Multicluster, Mobile, Multimedia Radio Network, Proc. ACM Wireless Networks, 1(3). 36. Iwata, A., C.C. Chiang, G. Pie, M. Gerla and T. Chen, 1999. Scalable Routing Strategies for Ad-Hoc Wireless Networks, IEEE Journal on Selected Areas in Communications, Special Issue on Ad-Hoc Networks, 17(8): 1369-1379. 37. Bellur, B., R.G. Ogier and F.L. Templin, (2004).Topology Dissemination Based on ReversePath Forwarding (TBRPF), IETF RFC 3684. 38. Lee, Y.Z., M. Gerla, J. Chen, B. Zhou and A. Caruso, 2006. Direction forwarding Routing, Proc. Ad-Hoc & Sensor Wireless Networks, 2(2). 39. Bertsekas, D.P. and R.G. Gallager, 1987. Distributed Asynchronous Bellman-Ford Algorithm, Data Networks, Prentice Hall, Englewood Cliffs, pp: 325-333. 40. Jun, J. and M.L. Sichitiu, 200). MRP: Wireless mesh networks routing protocol. Computer Communications, 31(7): 1413-1435. 41. Hamma, S., E. Cizeron, H. Issaka and J.P. Guedon, 2006. Performance evaluation of reactive and proactive routing protocol in IEEE 802.11 ad hoc network, Proc. SPIE 6387 Adhoc and Sensor Networks, Next-Generation Communication and Sensor Networks, 638709. 42. Liu, C. and S. Kaiser, 2005. A Survey of Mobile Ad-Hoc network Routing Protocols, Tech. Report and (2003-08). 885

42. Royer, E.M., 1999. A review of current routing protocols for ad-hoc mobile Wireless Networks, 6(2): 46-55. 43. Park, V.D. and M.S. Corson, 2001. TemporallyOrdered Routing Algorithm (TORA) version 4: Functional specification". Internet-Draft, draftietfmanet-TORA-spec-04.txt, July 2001. 44. Johnson, D. and D. Maltz, 1996. Dynamic Source Routing in Ad Hoc Networks. Mobile Computing, pp: 152-81. 45. Toh, C.K., 1997. Associativity-based routing for ad hoc mobile networks, Wireless Personal Communications, 4(2): 103-139. 46. Draves, R., J. Padhye and B. Zill, 2004. Routing in multi-radio, multi-hop wireless mesh networks. In Proc. of ACM MOBICOM.114-128. 47. Chakeres, I.D. and C.E. Perkins, 2010. Dynamic MANET On-demand (DYMO) Routing. http://tools.ietf.org/html/draft-ietf-manet-dymo-21. 48. Corson M.S. and A. Ephremides, 1995. A Distributed Routing Algorithm for Mobile Wireless Networks, ACM/Baltzer Wireless Networks, 1(1): 61-81. 49. Popa, L., A. Rostamizadeh, R.M. Karp, C. Papadimitriou and I. Stoica, 2007. Balancing Traffic Load in Wireless Networks with Curveball Routing, Proc. Eighth ACM International. 50. Aguayo, D., J. Bicket and R. Morris, 2003. SrcRR: A High-Throughput Routing Protocol for 802.11 Mesh Networks, http://pdos.csail.mit.edu/rtm/srcrrdraft.pdf. 51. Feng, J., R. Xia and H. Zhou, 2007. InterferenceAware Load Balanced Routing in Wireless Mesh Networks, Proc. International Conference Wireless Communications, Networking and Mobile Computing (WiCom'07), pp: 1730-1734. 52. Santivanez, C. and R. Ramanathan, 2003. Hazy sighted Link State (HSLS) Routing: A scalable Link State Algorithm, BBN technical memo BBN-TM-1301, BBN Technologies, Cambridge. 53. Hass, Z.J. and R. Pearlman, 1999. Zone routing protocol for ad-hoc networks, http://tools.ietf.org/html/draft-ietf-manet-zone-zrp-04. 54. Nikaein, N., C. Bonnet and N. Nikaein, 2001. HARP: Hybrid Ad-Hoc Routing Protocol, Proc. International Symposium on Telecommunications (IST'01). 55. Joa-Ng, M. and L.T. Lu, 1999. A peer-to-peer zonebased two-level link state routing for mobile Ad-Hoc networks, IEEE Journal on Selected Areas in Communications, 17(8): 1415-1425.

World Appl. Sci. J., 30 (7): 870-886, 2014

56. Hass, Z.J., R. Pearlman and P. Samar, 2002. The Intrazone Routing Protocol (IARP) for Ad-Hoc Networks, Internet Draft, http://www.ietf.org/ proceedings/ 02nov/I-D/draft ietf-manet-zone-iarp02.txt. 57. Siraj, M. and K.A. Bakar, 2012b. A Load balancing Interference Aware Routing Metric (LBIARM) for multi hop wireless Mesh Network Int. J. Phys. Sci., 7(3): 456-461. 58. Siraj, M. and K.A. Bakar, 2012. To minimize interference in multi hop wireless mesh networks using loadbalancing interference aware protocol, World Applied Sciences Journal, 18(9): 1271-1278. 59. De Couto, D.S.J., D. Aguayo, J. Bicket and R. Morris, 2003. A High-Throughput Path Metric for Multi-Hop Wireless Routing, in Proc. of the 9th annual international conference on Mobile Computing and Networking, pp: 134-146. 60. Draves, R., J. Padhye and B. Zill, 2004. Routing in multi radio, multi hop wireless mesh networks. In proceedings of ACM MOBICOM.114-128. 61. Yang, Y., J. Wang and R. Kravets, 2005. Designing routing metrics for mesh networks, Proc. WiMesh. 62. Yang, Y., J. Wang and R. Kravets, 2006. Interferenceaware loop-free routing for mesh networks. 64. Dijkstra, E.W., 1959. A Note on Two Problems in Connection with Graphs. Numerische Math., 1: 269-271. 65. Aiache, H., V. Conan, L. Lebrun and S. Rousseau, 2008. A load dependent metric for balancing Internet traffic in Wireless Mesh Networks," Mobile Ad Hoc and Sensor Systems, pp: 629-634. 66. Jiang, W., S. Liu, Y. Zhu and Z. Zhang, 2007. Optimizing Routing Metrics for Large-Scale MultiRadio Mesh Networks Wireless Communications, Networking and Mobile Computing, pp: 1550-1553. 67. Shila, D.M. and T. Anjali, 2007. Load-aware Traffic Engineering for Mesh Networks. Computer Communications and Networks, pp: 1040-1045. 68. Subramanian, A.P., H. Gupta, S.R. Das and J. Cao, 2008. Minimum Interference Channel Assignment in Multi-radio Wireless Mesh Networks, IEEE Transactions on Mobile Computing, 7(12): 1459-1473. 69. Pirzada, A.A., R. Wishart, M. Portmann and J. Indulska, 2009. ALARM: An Adaptive Load-Aware Routing Metric for Hybrid Wireless Mesh Networks, Proceedings of 32nd Australasian Computer Science Conference, pp: 25-34. 886

70. Eiman, A. and S. Roy, 2008. A Location-Aware Routing Metric (ALARM) for Multi-Hop, MultiChannel Wireless Mesh Networks, WCNC Proceedings, pp: 2081-2086. 71. Adya, A., P. Bahl, J. Padhye, A. Wolman and L. Zhou, 2004. A multi-radio unification protocol for IEEE 802.11 wireless networks, Proceedings of the First International Conference on Broadband Networks (BROADNETS04), pp: 344-354. 72. Awerbuch, B., D. Holmer and H. Rubens, 2006. The medium time metric: high throughput route selection in multi-rate ad hoc wireless networks, Mob. Networking. Appl., 11(2): 253-266. 73. Wei, Z., Z. Dongbo and D. Qiao, 2006. Comparative study of routing metrics for multi-radio multi-channel wireless networks, in Wireless Communications and Networking Conference, 1: 270-275. 74. Waharte, S., B. Ishibashi, R. Boutaba and D. Meddour, 2008. Interference-aware routing metric for improved load balancing in wireless mesh networks, in Proc. of 2008 IEEE International Conference on Communications, pp: 2979-2983. 75. Roy, S., D. Koutsonikolas, S. Das and Y.C. Hu, 2006. High-throughput multicast routing metrics in wireless mesh networks, in Proc. of 26 th IEEE International Conference on Distributed Computing Systems, pp: 48. 76. Zhao, X., C.T. Chou, J. Guo, S. Jha and A. Misra, 2008. Probabilistically reliable on demand multicast in wireless mesh networks, in Proc. of IEEE WoWMoM, 08: 1-9. 77. Park, J.C. and S.K. Kasera, 2005. Expected Data Rate: An Accurate High-Throughput Path Metric for Multi-Hop Wireless Routing, Proc. IEEE Communications Society Conference on Sensor and Ad-Hoc Communications and Networks. 78. Kyasanur, P. and N.H. Vaidya, 2006. Routing and Link-layer Protocols for Multi-channel MultiInterference Ad-Hoc Wireless Networks, Proc. ACM Special Interest Group on Mobility of Systems, Users, Data and Computing (SIGMOBILE'06). Mobile Computing and Communications, 10(1): 31-43.