Nash Equilibria Of Internet Congestion Control Game

The dissertation gives a summary of the research on congestion control in detail,then a universal game theoretical model for congestion control is proposed,the definition of Nash equilibria is given,the solutions to gain Nash equilibria are summarized.The existence of TCP game is discussed.A TCP game model is presented,in the game,the player is the network user that implemented by TCP algorithm,the strategy is the slow start congestion window parameter,each user can modify the parameter freely in order to gain more profit. Through mathematics analysis ,we conclude that there is a Nash equilibria in TCP game.Through Ns2 simulation we validate that there is a Nash equilibria in TCP game when the router used Drop Tail and end user implemented by Tahoe or Reno.A game model for Active Queue Management(AQM) and the condition for Nash Equilibria are also presented.In a AQM game, the player is the Possion traffic flows which implemented by UDP and the rule is AQM. We find that Drop Tail and RED do not impose a Nash Equilibria,CHOKe can impose a nearly Nash Equilibria,we modify CHOKe slightly to obtain a scheme that impose a Nash Equilibria.A new scheme that can impose Nash Equilibria is also presentd. At the same time,the conclusions are validated through simulations.The behaviors of congestion control based on infinite repeated game model are studied.Repeated and infinite repeated congestion control game models are presented,the discount factor in repeated game is discussed,the existence and optimization of the Nash equilibrium point is proved.Finally,we review the entire work and suggest the direction for future research.