Research On Channel Assignment Algorithm In Multi-radio Multi-channel Wireless Mesh Networks

Wireless Mesh Networks is a multi-hop wireless networks that consist of Meshrouters and clients, WMN is an important scheme to resolve the problem of "Last Mile"broadband access, because of those advantages, such as low up-front cost, easy networkmaintenance, robustness, reliable service coverage and high bandwidth and so on.However, the number of available non-overlapping channels is limited, so it is vitallyimportant that how to assign the channel. Wireless routers have typically been equippedwith multiple radios operating on non-overlapping channels to increase networkthroughput and improve the performance. In this thesis, the algorithm of channelassignment in multi-radio multi-channel WMN was studied. The achievements aresummarized as follow.Firstly, the feature of technology, application scenarios, and the key technology areanalyzed, and the standards used in WMN are introduced. Meanwhile, current channelassignment strategies are discussed and the objective and constraint requirements of CAare analyzed. And then, the advantage and disadvantage of current CA algorithms areconcluded, which lay a foundation for the research of later CA algorithms.Secondly, more precise physical model is adopted for interference estimating, asthat interference should be considered in the process of CA. With the optimization aimof maximizing the capacity, a CA model based on integer programming is proposed.The Lingo software is used to solve the CA models, and by setting different testscenarios, the validity and reliability of the model are verified, in the form of numericalanalysis. An approach based on greedy algorithm was proposed for reducing thecomplexity of CA. And the result of greedy algorithm is treat as initialized solution ofLingo Model, therefore the efficiency and quality was improved.Finally, a CA algorithm based on Discrete Particle Swarm Optimization isproposed. The mutation operation are is to change the position of particle in a randomlyway, and the crossover operation is used modify the current position of particle withinformation from the personal and global best, therefore, the algorithm of PSO isimproved to solve the CA problem of WMN. In addition, a process of channel mergingis added, that can eliminate the problem of radios constrain violated. And Matlab is usedto implement the algorithm, with simulations results showing that the scheme canreduce network interference effectively and improve the performance of network.