基于采样的gCHOKe主动队列管理算法的研究

Research on Improved gCHOKe Algorithm Based on Sampling

主动队列管理(Active Queue Management,AQM)算法是网络拥塞控制中非常重要的研究领域之一,文中主要研究了主动队列管理算法的公平性。gCHOKe算法是一种近似公平的主动队列管理算法,一旦发生gCHOKe击中则继续抽取分组进行比较直至无击中,但它仍然存在带宽公平性不足的问题,因此提高非适应性流的击中有效性以及惩罚非适应性流的力度是提高算法公平性的关键因素。文中在gCHOKe算法的基础上,提出了一种基于采样的主动队列管理算法sgCHOKe(Sampling based gCHOKe),在进行gCHOKe击中处理时首先采样多个分组与已到达分组进行比较,并采用新的策略机制丢包,来提高算法的公平性。仿真实验结果表明,sgCHOKe在维护网络带宽公平分配上性能优于CHOKe,gCHOKe算法,具有更高的击中有效性与带宽公平性,并且拥有更高的网络公平性指数。

Active Queue Management （ AQM） algorithm is one of very important research field of network congestion control. Focus on the fairness of the active queue management algorithm in this paper. gCHOKe is an algorithm of fair approximation of active queue man-agement. When gCHOKe hit happened, the algorithm extracts another packet and judge whether gCHOKe hit happening or not. The gCHOKe hit,whose effectiveness and the force of punishing non-response flows are the key factors to enhance the fairness algorithm,is lack of fairness. Based on gCHOKe algorithm,sgCHOKe （Sampling based gCHOKe） is proposed in this paper,through analyzing the low effectiveness of non-response flows hit,which samples several packets from queue to compare with the arrived packet and employ a new mechanism of packet-loss. The simulation results based on NS-2 show that compared with CHOKe and gCHOKe,sgCHOKe can maintain the fairness of different bandwidths better,achieving better effectiveness of non-response flows hit,with high index of network fairness.