一种基于博弈的拥塞控制改进算法G-Vegas

随着互联网的发展,网络拥塞问题越来越严重,如何改进现有的拥塞控制算法成为一个重要课题。为了解决网络拥塞问题,目前已有很多拥塞控制算法,大体可分为端到端的拥塞控制和基于网络的拥塞控制,本文主要关注基于端到端的拥塞控制。在众多的TCP拥塞控制算法中,Vegas算法以其主动避免拥塞的思想,具有较好的效果。但是,Vegas与目前主流的Reno算法兼容性差,存在带宽被挤占的问题。本文分析了拥塞问题的多重原因,并从博弈的角度分析了Vegas的缺点,提出了一种改进的拥塞控制算法G-Vegas。通过在NS2平台仿真,验证了算法的有效性。

An Advanced Congestion Control Algorithm G-Vega Based on Game Theory

With the development of the Internet,congestion has become more and more serious,and good congestion control algorithms are needed.Vegas is a good algorithm for its active avoidance of congestion,however,it can not work well with the mainstream algorithm Reno because its bandwidth can be stolen by Reno.This paper analyzes the problem between Vegas and Reno,with a method of game theory,and proposes an advanced congestion control algorithm G-Vegas.According to the results of emulation on NS2,the algorithm is effective.