Under The Framework Of Game Theory, Network Congestion Control And Bandwidth Allocation

With the rapid development of the Internet, the network resource especially bandwidth resource becomes scarce and network congestion is a fundamental problem today. However traditional approaches to congestion control and resource allocation ignore the QoS of different user and lack proper incentive-mechanism. By the introduction of game theory approaches, network users can independently optimize their utility functions and approach the Pareto optimization of resource allocation.Based on the theory of Nash equilibrium , sub-game perfect Nash equilibrium and Bayesian Nash equilibrium, we can draw some useful conclusion by the aid of theoretical analysis and network simulation.In the first place, we formulate a model of network resource allocation in non-cooperative game, giving a characterization of Nash equilibrium and Pareto optimization, analyzing the network "tragedy of the commons" and the benefit of congestion pricing.Secondly, we expound the PMP model, putting emphasis on the equilibrium and the effect of pricing mechanism that were not discussed in former reference. Founded on above analysis, network simulation under NS environment are conducted to evaluate performances of PMP.Thirdly, we carry out auction method in network engineering. We discuss the facticity of bid in virtue of Bayesian Nash equilibrium and theorems of mechanism design. Furthermore , the incentive compatibility of bid is also proved. Improving on PSP method, a novel assumption of auction is presented.Most research work in this field is based on static game. For the first time, we analysis the relation of network users in repeated games. According to the theory of chain—store paradox, folk theorem and KMRW reputation, we explain the network "tragedy of the commons" from another point of view and point out the strategy of congestion control under this condition.At last ,we review the entire work and suggest the direction for future research.