摘要
随着网络流量的剧增,主动队列管理是近来端到端拥塞控制研究中的一个研究热点。为了解决队列拥塞、高效利用队列资源和减少队列抖动等队列管理问题,采用模糊控制模块计算分组丢弃概率,减少了由于"硬判断"所造成的队列抖动问题;同时在算法中引入期望队列长度来显式控制并稳定队列长度,提高了队列资源利用率。最后通过在NS2平台上仿真三种不同的业务流量对该算法进行实验测试,结果表明,算法在一定程度上解决了PI算法队列抖动大和AVQ算法队列资源使用率低的问题,具有一定的可行性和应用价值。
With the dramatic growth of network traffics, AQM is becoming a focus in filed of TCP end - to - end congestion control research. In order to solve the problems of queue congestion, utilize queue＇s resource effectively and reduce queue＇s jitter, this text adopts a method of computing packet＇s drop probability through Fuzzy - Controlling Model to reduce queue＇s jitter caused by ＂hard -judging＂ ; at the same time, the expected queue length is introduced into the algorithm to explicitly control and stabilize the queue length, and improve queue＇s resource utility. Finally, three different kinds of traffics are simulated using NS2 simulation platform, the results of simulations demonstrate that the proposed algorithm is of a certain feasibility and application value , and can solve the problems of queue＇s jitter and low queue＇s resource utility to a certain extent.
出处
《计算机仿真》
CSCD
北大核心
2009年第10期111-114,118,共5页
Computer Simulation
