Research On Predictive Method Of Equilibrium Solution In Network Game Model

Entities in Internet,which can choose strategy independently,will change their strategies for better performance.Their behaviors will affect each other.One way to study the resource allocation problem under this scenario is to model the problem as a game and study the equilibrium.What this thesis did is to develop server selection methods under two different server selection scenes and a Nash equilibrium calculation method in a special scene by studying equilibrium prediction algorithm.In chapter one,this thesis introduces the research on equilibrium forecast,interaction between Internet Service Provider(ISP)and Content Provider(CP),and research on server selection problem.The research significance is proposed.In chapter two,this thesis proposes a server selection mechanism based on Wardrop Equilibrium prediction when users are edge routers in physical network.The simulation is done under game scene and dynamic scene.The results show that user delay is reduced by applying the server selection mechanism based on equilibrium prediction in high-linkutilization network.In chapter three,this thesis proposes a server selection mechanism based on correlated equilibrium prediction when users are independent computers in physical network.In order to reduce the delay of users,this thesis proposed an algorithm based on Genetic Algorithm to find a better traffic engineering solution.Experiments show that compared with the contrast scheme,the traffic engineering solution found by genetic algorithm can reduce the total delay of the user.In chapter four,chapter four,this thesis proposes a method to calculate the Nash equilibrium between ISP and CP based on relaxation algorithm.Simulation results show the scheme is only applicable to some scenes.The performance of ISP and CP when they reach Nash equilibrium is studied by simulation.The results show that compared with the utility when ISP and CP optimize their performance independently,the utility is better.In chapter five,summary of this thesis and further research is proposed.