首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种改进的显式拥塞指示算法
引用本文:赖峻,叶梧,冯穗力.一种改进的显式拥塞指示算法[J].华南理工大学学报(自然科学版),2008,36(1):38-43.
作者姓名:赖峻  叶梧  冯穗力
作者单位:华南理工大学,电子与信息学院,广东,广州,510640
摘    要:首先仿真分析了ECN(Explicit Congestion Notification)协议RFC 3168网关的性能.研究表明,该协议中建议的平均队列长度超过门限上界时丢弃包的策略将导致网络吞吐量过低.如果改为标记包策略则可以显著提高吞吐量,但因此造成的长队列会增加系统的平均延迟,而且当负载持续增加时还可能导致不能成功建立连接。为保证高吞吐量、低延迟和连通的鲁棒性,提出一种名为LQRD(Long Queue Random Drop)的改进算法,该算法当平均队列超过队列上限时随机的丢弃分组.丢弃概率综合考虑到平均队列、路由器的缓存大小、链路容量和系统平均延迟等因素的影响,仿真结果表明其具有良好的性能.最后,我们给出并分析了该算法的解析模型.

关 键 词:拥塞控制  显式拥塞指示  RED  
文章编号:1000-565X(2008)01-0038-06
收稿时间:2007-01-26
修稿时间:2007年1月26日

An Improved Explicit Congestion Notification Algorithm
Lai Jun,Ye Wu,Feng Sui-li.An Improved Explicit Congestion Notification Algorithm[J].Journal of South China University of Technology(Natural Science Edition),2008,36(1):38-43.
Authors:Lai Jun  Ye Wu  Feng Sui-li
Abstract:The performance of gateways defined by RFC3168, which is ECN (Explicit Congestion Notification) protocol, is analyzed firstly by simulating. The study shows that the recommendation in the protocol that gateways drop packets when the average queue size exceeds the upper threshold brings too low throughput. However, if it is be substituted for marking packets, which increases average delay and induces parts of fail connections for persistent heavy loads, the throughput will improve greatly. To make sure high throughput, low average delay and robust connectivity for long queue length, an improved algorithm named LQRD (Long Queue Random Drop) is presented. The algorithm drop packets randomly when the average queue size exceeds the upper threshold. The drop probability considers average queue length, buffer size, link capacity and average queue delay synthetically. Simulation shows the algorithm has good performance. And at last, we give and analyze the solved model of the LQRD algorithm.
Keywords:gateway  congestion control  explicit congestion notification  random early detection
本文献已被 维普 万方数据 等数据库收录!
点击此处可从《华南理工大学学报(自然科学版)》浏览原始摘要信息
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号