自相似业务RED的新型算法分析  

Analysis of a novel predictable RED Algorithm with Self-Similar Traffic Input

作　　者：黄媛辉[1] 谭献海[1] 刘黎娜[1] 胡英[1] 

机构地区：[1]西南交通大学信息科学与技术学院,四川成都610031

出　　处：《微计算机信息》2008年第21期153-155,共3页Control & Automation

基　　金：国家自然科学资金项目(90104002)

摘　　要：本文提出一种自相似业务下适合IP网络拥塞控制的新型算法SSRED(Self-Similar traffic Random Early Detection)。将自相似参数H引入权值函数,通过平均队长和队长方差推导最大与最小队长阈值,最后根据Norros所提缓冲区溢出概率公式修改最大丢包率。仿真证明这种算法在自相似业务下能够提高利用率,减少报文丢失率和网络时延等性能。In this paper, an attempt has been made by proposing a novel predictable RED Algorithm （SSRED） for traffic congestion management in IP networks with self-similar input. The self-similar parameter H had been introduced into the weight function, and we had also derived the thresholds of the maximum and minimum queue size based on self-similar traffic modelling. Finally we had also modified the maximum packet dropping probability of the proposed algorithm according to buffer overflow probability that Norros had proposed. Simulation results demonstrate that this algorithm can be used to improve the utiliazation, reduce the packet loss rate and the network delay under the self-similar traffic.

关 键 词：自相似业务 SSRED 权值函数 丢包率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]