Game-theory-based Network Selection Algorithm In Heterogeneous Wireless Networks

Due to the highly integration of network, the research on resource management architecture for autonomous and heterogeneous network is important and demanding. However, the study on this topic is still on its early stage. As the game theory has been utilized to investigate the heterogeneous network, it is promising to apply the latest and novel results of game theory for developing the resource management architecture for autonomous and heterogeneous network.In the third section, the evolution game algorithm for network selection in heterogeneous network environment is investigated. A heterogeneous network model consisting of active clients is established according to the network-driven based network selection algorithm. The formula for unit price and transmission rate of two pricing schemes, the uniform pricing and differentiated pricing, are introduced and derived in the model. According to the analysis on Niyato's evolutionary game approach, an improved evolutionary game algorithm is proposed. To evaluate the performance of the improved evolutionary game method, simulations on both the original and proposed evolutionary game algorithms are performed. The simulation results indicate, comparing to Niyato's algorithm, the proposed algorithm has higher aggregate payoffs and unnecessary handovers while a larger average number of iterations needed for convergence. Since that the average number of iterations needed for convergence is not a critical issue when the number of users in the network is not very large, the proposed improved evolutionary game approach has a relatively good performance.In the fourth section, Chinese restaurant process is utilized to solve the network externality. A wireless network model is built based on definition of network externality, and the recursive form of utility function for the model is derived through best response strategy based on Chinese Restaurant Game. To compare the performances of the best response strategy with four existed methods, myopic strategy, learning strategy, signal strategy and random strategy, simulations are performed. The simulation results demonstrate that, the best response strategy derived from Chinese Restaurant Game has a better performance than other four strategies.