华南理工大学学报(自然科学版) ›› 2009, Vol. 37 ›› Issue (5): 90-93,105.

• 计算机科学与技术 • 上一篇    下一篇

基于网络演算的有损分形整形器及其性能模型

张连明1  刘孙东2   

  1. 1. 华南理工大学 计算机科学与工程学院, 广东 广州 510006;2. 深圳信息职业技术学院 电子通信技术系, 广东 深圳 518029
  • 收稿日期:2008-07-10 修回日期:2008-09-28 出版日期:2009-05-25 发布日期:2009-05-25
  • 通信作者: 张连明(1972-),男,博士后,湖南师范大学副教授,主要从事计算机网络和复杂网络研究. E-mail:zhanghn@scut.edu.cn
  • 作者简介:张连明(1972-),男,博士后,湖南师范大学副教授,主要从事计算机网络和复杂网络研究.
  • 基金资助:

    国家自然科学基金资助项目(60773012);中国博士后科学基金资助项目(20070420782);湖南省自然科学基金资助项目(07JJ6127)

Loss Fractal Regulator Based on Network Calculus and Its Performance Model

Zhang Lian-ming1  Liu Sun-dong2   

  1. 1 School of Computer Science and Engineering, South China University of Technology, Guangzhou 510006, Guangdong, 2. Department of Electronics Communication Technology, Shenzhen Institute of Information Technology, Shenzhen 518029, Guangdong, China
  • Received:2008-07-10 Revised:2008-09-28 Online:2009-05-25 Published:2009-05-25
  • Contact: 张连明(1972-),男,博士后,湖南师范大学副教授,主要从事计算机网络和复杂网络研究. E-mail:zhanghn@scut.edu.cn
  • About author:张连明(1972-),男,博士后,湖南师范大学副教授,主要从事计算机网络和复杂网络研究.
  • Supported by:

    国家自然科学基金资助项目(60773012);中国博士后科学基金资助项目(20070420782);湖南省自然科学基金资助项目(07JJ6127)

摘要: 为改善Internet流量自相似等突发特征给网络性能带来的影响,提出了一种适合自相似突发流量的有损分形整形器的数学模型,分析了有损分形整形器的队列长度与延迟等性能,讨论了有损分形整形器的引入对端到端延迟、数据丢失总数以及平均包丢失率等的影响,得到了各性能指标上界与流量自相似参数的关系,给出了允许最大平均包丢失率下的最小缓存容量.这对基于分形整形器的拥塞和流量控制策略的评价、分形整形器自身参数的设计具有参考作用.

关键词: 自相似流量, 有损分形整形器, 确定性网络演算, 模型, 上界

Abstract:

In order to avoid the network performance degradation due to the burst characteristics including Internet self-similar traffic, etc. , a mathematical model of loss fractal regulator controlling the self-similar traffic is pro- posed. The performances of the regulator including the queue length and the delay are analyzed, and the influences of the introduction of the regulator on the end-to-end delay, the total packets loss and the average loss rate are dis- cussed. Then, the relationship between the upper bounds of the above-mentioned performance indexes and the pa- rameter of self-similar traffic are investigated. Moreover, the buffer minimum of the loss fractal regulator at the maxi- mum average loss rate is obtained. The above-mentioned results provide a useful reference to the evaluation of con- gestion control scheme and traffic management scheme of self-similar traffic and to the design of parameters of the loss fractal regulator

Key words: self-similar traffic, loss fractal regulator, deterministic network calculus, model, upper bound