Research On Capacity And Cross-layer Design In Wireless Mesh Network

With the development of multi-hop and heterogeneity techniques, wireless mesh network has emerged as a key technology for next generation wireless network. Wireless mesh network is dynamically self-organized and self-configured, with the nodes in the network automatically establishing an ad hoc network and maintaining the mesh connectivity. This new type of networks has solved many problems that exist in the traditional wireless networks, e.g., lack of expansibility and robustness. However, these novel and powerful techniques also change the way of utilizing wireless resource, and further become a challenge for the traditional layered resource allocation algorithm. In fact, significant performance gains can be achieved by various cross-layer approaches in this network, and it is necessary for the future design of wireless network protocol stack. In this dissertation, under the guidance of information theory, network theory and convex optimization, the capacity and cross-layer design in wireless mesh network are studied.Firstly, the cross-layer optimization problem of maximizing the capacity of a regular multi-radio, multi-channel wireless mesh network was addressed. Given certain network topology and traffic demands, by jointly considering power control, channel assignment and scheduling, studied how to maximize the network capacity under fairness constraint. Based on graph theory, the problem was divided into several sub-problems, which can be formulated as linear programming. It was proved that the solutions of these sub-problems will lead to the global optimal solution to our problem. However, the computational complexity of the algorithm was extremely high, as it needed to traverse all possible scenarios. Therefore, a suboptimal algorithm was proposed, which only required solving a local optimization sub-problem in the bottleneck area. The suboptimal algorithm achieved a very low computational complexity at the cost of moderate performance degradation.Next, a cross-layer optimization problem for multiple real-time video traffics in wireless mesh networks was addressed. Given certain network topology, real-time video traffic demands and their routings, by jointly adjust the parameters of the transport layer and application layer, studied how to minimize the overall video distortion under strict QoS constraints. The QoS constraints were modeled in this dissertation as a three-dimensional variable: rate, delay and packet loss rate, and the performances of application and transport layers were modeled by some classical models under given assumptions. Then, a framework was presented, in which source coding, packet loss rate, ARQ control, and FEC functionalities at different layers can be jointly optimized. Thereafter, a centralized algorithm was proposed, and the cross-layer optimization problem was divided into several sub-problems, which can be formulated as convex optimization problem and can be solved efficiently by some classical convex programming methods. Base on the previous study, an admission control algorithm was also proposed to judge whether a real-time video traffic was able to access to the wireless mesh network.Since there were no central control nodes in wireless mesh network, centralized algorithm was not applicable. Therefore, in the last part of the dissertation, distributed protocol design was addressed. Based on the Lagrangian dual decomposition technique, the cross-layer optimization problem for multiple real-time video traffics was divided into three sub-problems: source coding rate sub-problem, delay partition sub-problem and packet loss rate sub-problem. A distributed algorithm was proposed to solve the first two sub-problems, and by using the Newton gradient method, the third sub-problem was solved iteratively to approach the optimal solution. Based on the previous algorithm, a distributed and applicable protocol was designed.