Clustering And Cross-layer Cooperation-based Performance Optimization In Wireless Ad Hoc Networks

This thesis is concerned with wireless ad hoc networks, which has become one of the most favorite area in wireless mobile communication. By analyzing the exiting problem of wireless ad hoc networks, this thesis mainly studies framework, minimal connected dominating sets(MCDS) algorithms and TCP performance, which aims at the performance optimization in wireless ad hoc networks. Firstly, to deal with a series of problems, which are caused by infrastructure layering in wireless node, a mobile agent based open framework for wireless ad hoc networks is proposed with the help of mobile agent and active networking technologies. Secondly, a new algorithm of minimal connected dominating sets in wireless ad hoc networks is proposed by using coloring algorithm in graph theory. The MCDS generates the virtual backbone for wireless ad hoc network. The performance of the MCDS is proved by theoretical analysis and simulation, which show that the MCDS algorithm works well especially in dense environment. Finally, the existing problem and current development of TCP performance in wireless ad hoc networks was introduced. Based on the cooperation of TCP and route layers, a new strategy to enhance TCP performance, called RP-DSR, is proposed and applied to dynamic source protocol(DSR) in NS2 simulator. The simulation result shows that the RP-DSR outperforms DSR.