Research Of Capacity Optimizing Model And Algorithm On Wireless Mesh Networks

As a new type of wireless network infrastructure, because of its self-organization, Wireless Mesh Networks have been paid extensive attention both in international academia and industry. As a basis issue for the wireless Mesh network, network capacity is the most important indicator to determine the network performance, which relates to the network topology, routing algorithm and channel allocation, load balancing and other aspects. The research goal of this paper is to achieve actual optimization network performance as much as possible theoretical optimum network capacity by a variety of strategies and methods such as optimizing the network topology, using the optimum channel allocation ways and routing algorithm and load balancing so on. The study of this paper includes several topics as the follow:(1) The relationship between network topology,network nodes,gateway number,channel number,interface number and network capacity are independent, and the relationship has been revealed by proof and derivation. Theoretical analysis shows that the performance of Grid networks is more effective than random networks, and only when the number of AP and Gateway,interface number and the number of available channels maintain a certain reasonable relationship, the networks can achieve theoretical optimum network capacity.(2) Deployment program of AP in wireless Mesh networks has a great impact on network performance, and the question has been studied in this paper. Random coverage ways and grid deployment ways have been proposed, and the performance compared by a statistical method based on Monte Carlo method, also two common AP deployment programs cluster-based and ring-based have been introduced.(3) The deployment program of gateway has been studied in this paper. In the paper three optimization objectives as minimizing the number of gateways, minimizing the average number of hops from AP to gateway, and minimizing load-balancing index were proposed, and to achieve the three objectives, a heuristic solution based on greedy algorithm and clustering algorithm has been presented. In the solution, an indicator parameter reflecting the number of gateways and hops was designed, and a good initial approximate solution can be obtained by a greedy algorithm based on the indicator parameter, then initial approximate solution was adjusted and optimized to minimize the load balance index and the network can be distributed symmetrically. Experiment results have shown that the three optimization objectives can get a good balance status by the proposed solution.(4) Channel allocation and routing algorithm are two important topics in Wireless Mesh networks, the two topics are interdependent and affect each other and play a very important role on the network performance. Channel allocation and routing algorithms has been studied jointly in this paper. To improve the efficiency of channel assignment and routing, a network topology optimization algorithm based on maximum flow minimum cut was proposed in first. And a joint channel allocation and routing solution was proposed based on avoiding interference and topology optimization. In the solution, based on topology optimized, the distribution of traffic of each link can be calculated by a routing algorithm, then each link was grouped under the constraints of link utilization and node interface, at last each group was assigned a channel by the principle of minimizing interference. Experiments show that the proposed solution can avoid channel allocation chain reaction and can achieve fairly good performance and throughput.(5) Load balancing of data flow has been studied in this paper and a data flow load-balancing method was proposed based on maximum surplus flow spanning tree algorithm. For each network unit, first flow of each link was calculated by maximum flow algorithm, and then a maximum surplus flow spanning tree was built by cutting based on flow information. So the multi-path routing was transformed into a single-path routing by the spanning tree, and the original flow was scaled on each path. The proposed algorithm can maintain the balance between the fairness of each user and maximizing network throughput.The network capacity and performance can be enhanced by the research above, and the research has certain significance to network engineering practice.