随机早期检测RED算法的分析及改进  

Analysis and Improvement of Random Early Detection Arithmetic

作　　者：杜玉林[1] 杨海澜[1] 孙远伟[1] 

机构地区：[1]武汉交通职业学院电子与计算机系,湖北武汉430065

出　　处：《重庆电力高等专科学校学报》2009年第1期37-39,47,共4页Journal of Chongqing Electric Power College

摘　　要：随机早期检测RED(Random Early Detection)算法是广泛应用于路由器中的IP层的拥塞控制算法之一,它的主要思想是在拥塞发生以前,通过计算队列中包的丢失概率,从而随机丢弃一部分数据包,以达到实现网络拥塞控制的目的。但该算法在应用中仍有不足。针对于此,重新计算在缓冲区最大门限值附近的丢弃概率,提出新的改进算法,以增大网络吞吐量。The RED （Random Early Detection） algorithm, which is widely used in routers, is one of the congestion control algorithm in the IP layer. The main idea is to discard randomly a part of the packets by calculating packet＇ s drop probability before the congestion, so as to stop the network congestion. But the algorithm has some shortcomings in the applications. Therefore abandoned probability is calculated afresh near the maximum value in the buffer and a new algorithm is proposed to increase the throughput of the network.

关 键 词：随即早期检测 丢弃概率 拥塞控制 平均队列长度 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]