Multi-hop Wireless Network Based On The Agent Channel Assignment And Routing Algorithms

Multi-radio multi-channel wireless multi-hop technology utilizes all of the available channels to increase network capacity, which becomes a hot research issue of the next generation wireless broadband access. This dissertation first analyzes the challenges brought by multi-radio multi-channel. Then aims of the tremendous demand of network capacity for users, agent theory, negotiation and learning mechanism which belong to artificial intelligence are introduced to design channel assignment and routing algorithms.A distributed channel assignment algorithm based on multi-agent is proposed. The assignment of network's channel resource is achieved through negotiation and decision-making among router node agents. Response agent is designed to generate a competitive value according to its channel, radio and about its neighbors. Management agent is designed to negotiate with response agents by multi-issue metric, then the response agent with the optimal competitive value will be selected. At the same time, gateway agent is adopted to assess the whole system, so that to realize global optimization. A utility function is designed to evaluate the benefit of any radio-channel pair. Finally the neighbor-radio binding algorithm and the radio-channel binding algorithm are designed.A routing algorithm based on affinity and emotional learning is proposed to solve the problem of real-time updating of routing metric in source initiated on-demand routing protocol and the problem of selecting the dated optimal routing which caused by the numerous of information. Factors which influence the routing selection are designed to be affinity between optimal routing and candidate routing. And emotional learning method is introduced. According to network information obtain by mobile agents and agent's emotion, the coefficient of affinity factors can be updated locally or globally, so that the optimal routing can be obtained.Finally, the NS2 simulation tool is expanded to make it applicable to multi-radio multi-channel environment. Several different scenarios are designed to evaluate channel assignment routing algorithm. And the improvement of throughput, delay, routing load to the whole network is analyzed.