基于自相似流量预测的TCP拥塞控制算法  

TCP congestion control algorithm based on self-similar flow prediction

作　　者：孙珍[1] 李世银[1] 王秀娟[1] 徐京 

机构地区：[1]中国矿业大学信电学院,江苏徐州221116 [2]兖矿集团菏泽能化有限公司,山东菏泽274000

出　　处：《计算机工程与设计》2010年第12期2713-2715,2748,共4页Computer Engineering and Design

摘　　要：针对高带宽网络环境下传统TCPReno的拥塞控制效率低的问题,提出了一种适用于随机丢失网络的TCP拥塞控制改进算法。介绍并分析了网络流量自相似特性和预测的时间尺度,并进行了TCP链路流量预测及可用带宽估计,当网络出现丢包时根据估计的带宽动态设置慢启动阈值。OPNET仿真结果表明,该算法在高随机丢包的情况下吞吐量得到了明显改善。In high bandwidth network,an important challenge is to improve congestion control efficiency of TCP Reno.A new TCP congestion control algorithm which is suit for random packet loss network is presented.A survey on self-similarity and predicted time scale is given.TCP link’s flow prediction and available bandwidth estimation are held.Slow start threshold is set dynamically by available bandwidth when high packet loss occurs.Simulation results show that network performance is improved evidently in such en-vironment.

关 键 词：自相似 流量预测 带宽估计 拥塞控制 OPNET 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]