Research On Transmission Performance And Cross-layer Joint Optimization Algorithm In Wireless Mesh Networks

Wireless Mesh Network is a new kind of distributed networks with high-capacity and high-speed. It provides mobile users with the access to the backbone networks within its multi-hop wireless mesh structure. Comparing with traditional wireless networks, it has a promising future for various applications and has advantages such as high spectrum efficiency, coverage ability, good compatibility and low deployment cost, etc. Due to the variety of its service mode and the complexity of its structure as well as protocols, traditional network analysis methods and layered optimization algorithms can hardly meet its design requirements. This dissertation tries to analyze the transmission performance of wireless mesh networks and therefore to design a channel-routing joint optimization algorithm in order to improve the network capacity.To begin with, this dissertation makes an introduction to the fundamental concepts of wireless mesh networks. It also covers the current research progress, as well as the key problems and typical models, of transmission performance analysis and cross-layer joint optimization.In the analysis of modeling the transmission performance, we find that the main difficulties are the complex network structure and diversified communication technologies and protocols, which lead to the inapplicability of traditional unified modeling method based on single hop and IEEE 802.11 MAC protocol. For the purpose of distinguishing the impacts of different routing and MAC protocols and extending the applicability, this dissertation proposes a two-layered analysis framework including two sub-models: network model and node's channel model. Network model, based on queuing theory, considers network topology, data flow characteristics and multipath routing; while the node's channel model considers multi-channel and multi-hop features. Moreover, with regard to the node's average time for successfully accessing the channel and the average waiting time in the queue, this dissertation deduces the mathematic formulas for average network delay and throughput. In addition, In order to evaluate the accuracy of this framework, this dissertation calculates the average network delay under scenarios with TDMA MAC protocol and 802.11 MAC protocol respectively (both of them use AODV routing protocol). The results of NS2 simulation show that this framework could accurately estimate the transmission delay performance.In the analysis of cross-layer joint optimization, as opposed to the traditional single-layer optimization method which is incapable of utilizing new features of wireless mesh networks, this dissertation analyzes the bottleneck factors based on the transmission performance model and proposes a channel-routing joint optimization algorithm with feedback mechanism. This algorithm consists of heuristic-based search algorithm, primal-dual-based optimal path selection algorithm and channel assignment adjustment algorithm. The heuristic search algorithm uses the"potential residual bandwidth"to effectively reduce the channel collision; the optimal path selection algorithm belongs to the global optimization method; according to the monotonicity and feasibility of solutions for sub-problem, time control module is integrated into this algorithm hence the running time is effectively reduced. In the end, this dissertation uses Matlab to evaluate this algorithm under consideration of randomized topology, fixed number of channels, limited bandwidth as well as data rate, number of radios and iterations in time control, etc. Results show that this channel-routing joint optimization algorithm could successfully improve the overall network throughput.