首页 | 本学科首页   官方微博 | 高级检索  
     

典型AQM算法的性能评价模型
引用本文:汪浩,严伟.典型AQM算法的性能评价模型[J].计算机学报,2006,29(4):552-560.
作者姓名:汪浩  严伟
作者单位:北京大学信息科学技术学院网络实验室,北京,100871;江西师范大学软件学院,南昌,330027;北京大学信息科学技术学院网络实验室,北京,100871
基金项目:中国科学院资助项目;科技部科研项目
摘    要:利用GI/M/1/N排队系统和Internet业务流量自相似性的特点建立了一个评价AQM算法在非响应业务流量下性能的分析模型,提出了利用模型的分析计算结果而不是模拟或实验手段评价AQM算法性能的新方法.同模拟或实验手段相比,该方法能更深刻地刻画AQM算法在实际网络环境中的性能.用该模型分析比较3个经典的AQM算法——TD、RED和GRED的性能,所得的结果同其他研究者利用模拟或实验方法所得的结果一致.

关 键 词:RED算法  主动队列管理(AQM)  拥塞控制  GI/M/1/N排队系统  重尾分布  自相似网络流量
收稿时间:2004-07-07
修稿时间:2004-07-072005-11-24

A Performance-Evaluation Model of Typical AQM Algorithms
WANG Hao,YAN Wei.A Performance-Evaluation Model of Typical AQM Algorithms[J].Chinese Journal of Computers,2006,29(4):552-560.
Authors:WANG Hao  YAN Wei
Affiliation:1.Network Laboratory, School of Electronics Engineering and Computer Science, Peking University, Beijing 100871;2.School of Software, Jiangxi Normal University, Nanchang 330027
Abstract:The Internet traffic consists of responsive long-lived TCP flows and unresponsive short-lived TCP flows and UDP flows.A lot of papers have been published to model the interaction between long-lived TCP flows and AQM algorithms.However,unresponsive flows dominate the Internet traffic,which contribute 70%~80% of the Internet traffic.Therefore,it is important to model the behavior of AQM algorithms with unresponsive flows.In this paper,an analytical model is presented based on the GI/M/1/N queuing system and the self-similar traffic of the Internet to evaluate the performance of AQM algorithms with unresponsive flows.Compared with the presented experimental or simulation approaches,this modeling approach can help to reveal the underlying characteristics of the performance of AQM algorithms in a practical network environment.Using this model,the authors analyze the performance of three AQM algorithms: TD,RED and GRED.
Keywords:RED algorithm  active queue management(AQM)  congestion control  GI/M/1/N  heavy-tailed distributions  self-similar network traffic
本文献已被 CNKI 维普 万方数据 等数据库收录!
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