一种TCP博弈模型的Nash均衡存在性分析与仿真

Analysis and Simulation to Existence of Nash Equilibria in a TCP Game

在当今的Internet中,远程教学、大规模传输等"不响应拥塞"应用与日剧增,使得端系统有动机更改拥塞控制方案以满足自己的需要,这加重了Internet拥塞。一般希望通过研究用户非合作博弈行为的Nash均衡来规范自私的端用户行为。阐述了TCP拥塞博弈模型,主体为采取TCP协议的端用户,策略为TCP端用户的慢启动拥塞窗口递增参数。通过数学分析方法论证了网络方对TCP流没有额外的处罚时,TCP博弈存在Nash均衡。同时,通过Ns2仿真工具验证了当路由器采用Drop Tail队列管理算法,端节点采取TCP Tahoe和TCP Reno时,TCP博弈存在Nash均衡。这意味着TCP算法对Internet的稳定起着重要作用。

In current Intemet, the uncongestion reaction applications such as distance education,large ＂scale transmissions are increasing , so the users have incentives to modify its congestion strategy to satisfy their own needs, this worsen Internet congestion situation. In order to control the selfish behaviors, Nash equilibrium theory of uncooperative game is studied. In this paper, a TCP game model is presented, the player is the network user that implemented by TCP algorithm, strategy is the slow start congestion window parameter. Through mathematics analysis, conclude that there is a Nash equilibrium in a TCP game when no punishments are given to the TCP flows. At the same time, through Ns2 simulation, validate that there is a Nash equilibrium in a TCP game when the router used Drop Tail and end user implemented by Tahoe or Reno. This means that TCP congestion algorithm play an important role in Internet stability.