On Throughput Capacity Analysis And Channel Assignment In Wireless Mesh Networks

As one kind of multi-hop transmission wireless network, wireless mesh network (WMN) gains prosperous development and widely used in different applications. Wireless mesh network is originally derived from the traditional wireless mobile Ad Hoc networks. On one side, both wireless Ad Hoc network and WMN share some similarities, such as both of them do support multi-hop wireless signal transmission which means that each wireless node not only send its own data but also have to relaying the data traffic originated from those nodes that are far away from the central gateway node. Therefore, the coverage of the whole network is enlarged and the scalability is ensured. On the other side, WMN is obviously different with wireless Ad Hoc networks. The differences mainly concerned with the classification and mobility of wireless nodes. As far as wireless Ad Hoc network is concerned, the nodes in it are almost similar with each other because all nodes are highly mobile. Consequently, the research on wireless Ad Hoc network mainly focuses on dealing with dynamic topology management and subsequent problems, such as routing, power saving and quality of service. When it comes to WMN, based on the different functionality, the nodes are classified into two different kinds:Mesh routers and mesh clients. Mesh routers are used to build the backbone network architecture and have no mobility and free from power constraint. Mesh routers are responsible for providing major network service. Mesh routers, not only have the ability to connected with the Internet by equipped mesh routers with bridge or gateway functionalities, but also can be used to interconnect with different kinds of wireless networks and, thus, distinguished WMN with its high scalability. Mesh clients, on the contrary, can be any kinds of wireless devices and are usually mobile. Thus, the research area and future direction on wireless Ad Hoc network and WMN are quite different.There are three different kinds of WMN network models:Infrastructure/Backbone WMN, Clients WMN and Hybrid WMN. Among three of them, Clients WMN, mainly composed by traditional wireless mobile devices with mesh, is not representative. Both infrastructure/backbone WMN and hybrid WMN are mainly composed by mesh routers to form the backbone of the whole network. The backbone architecture is crucial to the network performance for it is not only responsible for connecting with the Internet, but also provides a way of interconnecting with other kinds of wireless networks. Consequently, the way of improving the performance of the mesh routers plays an important role in WMN network performance upgrading. Traditional mesh routers are equipped with single channel single radio interfaces (SCSR) in which the throughput capacity of each node are dramatically declined due to severe mutual interference among multiple simultaneous transmission signals when the total number of mesh routers deployed increased. Aimed at dealing with the throughput capacity degradation problem in SCSR wireless mesh networks, a great deal of resolutions are proposed, such as multi-channel multi-radio interface (MCMR), directional antennas, multiple input multiple output (MIMO) technology, cognitive radio technology and cross layer design and so on. Among them, multi-channel multi radio technology is widely used with the hope of improving throughput capacity performance. Some channel assignment algorithms are also developed in order to obtain optimal throughput capacity improvement performance.By applying multi-channel multi-radio technology in WMN, this paper gives deep research and focuses on the following area:Firstly, the node localization algorithms used in the traditional wireless networks are introduced in this paper to compute and evaluate the distance between any pairs of wireless nodes. The experiments are carried out by install real equipments and testbed. A great volume of data on the relationship between the value of the received signal strength indicator (RSSI) and the distance of any two nodes are collected. By applying the curve fitting method on those data, a precise computation formula to compute the distance based on the RSSI is proposed. This formula is useful in modeling the interference between transmitting signals, the interference range and transmission range. The distance computation formula lays the foundation of our following research. The interference between concurrent transmitting signals is classified into different categories. Then, the interference models, including capture threshold model, protocol interference model and physical interference model, are built based on the distance computation formula. Among them, the capture threshold model and protocol interference model are defined based on the distance between the concurrent transmitting nodes and, consequently, the interference degree is highly affected by the precision of the RSSI-based distance computation formula. The physical interference model, on the contrary, is mainly built based on the signal-to-interference (SIR) or signal-to-interference noise ratio (SINR) to model the mutual interference among multiple concurrent transmissions. Both the disadvantages and advantages of each kinds of interference model are concluded following which the minimum interference range is used to compare the interference mode proposed in this paper. Next, by using distance computation method, the total number of relaying hops that is needed to reach the central gateway node is computed. The nodes in the WMN networks are organized based on their hop distance from the gateway node, the whole network coverage area and network size. By doing this, different WMN network topology architectures are built, such as chain topology model, grid topology model, tree-based topology model and ring-based topology. The difference and exchange method among different architectures are concluded to help the throughput capacity analysis and channel assignment research in WMN with MCMR.Secondly, the complexity of applying three kinds of interference model into the process of throughput capacity analysis is compared and the protocol interference model is chosen to model the mutual interference among concurrent transmission links for its precision and low complexity. As network topology is concerned, the ring-based with the gateway node centered at the whole coverage area is chosen. Based on the asymptotic analysis method, the following conclusions are made:in WMN with single gateway node, the throughput capacity is shown to be inversely related with the total number of wireless nodes, the total rings in ring-based network topology, the ratio between the number of radio interfaces equipped in each mesh nodes and the channel that can be used. The analysis is then extended into WMN network with multiple gateway nodes, and the result shows that the throughput capacity is proportional to the total number of radio interface installed on each mesh nodes and the sum of the gateway nodes, while inversely related with the total number of channels that can be used and the number of wireless nodes.Finally, in order to deal with the drawback of simple assumption on channel assignment in the process of throughput capacity analysis, the channel assignment algorithms based on the linear programming theory in WMN with MCMR are proposed and evaluated. The problem of channel assignment in WMN with multi-channel multi-radio is modeled as the linear programming formulation. The channel assignment problem is equivalent with the objective of minimizing the network interference, while the network connection is preserved, and is constraint by the network interference constraint, connectivity degree and interface number constraints. Then, as the data traffic load on each node have great impact on the extent of the mutual interference, the linear programming-based channel assignment algorithm is extended and proposed the data traffic-aware channel assignment algorithm. In data traffic-aware channel assignment algorithm, the objective function is not only reflects the degree of mutual interference, but also shows the role of data traffic on nodes that are related with the concurrent interference transmissions. The channel assignment algorithms proposed are evaluated by using NS-2 simulator and deployed in a course-ware management system. The result shows that the throughput capacity degradation, due to enlargement of network size, varying of available channels and radio interfaces, is mitigated on some extent by using effective channel assignment.