作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2006, Vol. 32 ›› Issue (23): 13-14. doi: 10.3969/j.issn.1000-3428.2006.23.005

• 博士论文 • 上一篇    下一篇

基于FARIMA模型的网络排队性能分析

饶云华1,曹 阳1,杨 艳1,王习稳2   

  1. (1. 武汉大学电子信息学院,武汉 430079;2. 国营武昌造船厂,武汉 430060)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2006-12-05 发布日期:2006-12-05

Performance Analysis of Network Queue Based on FARIMA Model

RAO Yunhua1, CAO Yang1, YANG Yan1, WANG Xiwen2   

  1. (1. School of Electronics Information, Wuhan University, Wuhan 430079; 2. Satate Wuchang Shipyard, Wuhan 430060)
  • Received:1900-01-01 Revised:1900-01-01 Online:2006-12-05 Published:2006-12-05

摘要: 利用能够同时反映通信量长程和短程相关性的FARIMA模型,研究了长程相关和短程相关对于FIFO排队系统性能的影响,讨论了在给定系统缓存溢出概率条件下,FARIMA模型为输入时,排队队长分布的渐近解析表达式。研究表明,短程相关性在缓存较小时,对系统队长分布会有影响,但是在自相似存在的情况下,系统的排队长度分布为渐近Weibull分布,与通信量的短程相关性无关。蒙特卡罗仿真分析表明了结果的正确性和有效性。

关键词: 队长分布, 溢出概率, FARIMA, 长程相关

Abstract: The FARIMA traffic model which holds long-range dependence and short-range dependence simultaneously is employed to study the performance of FIFO queue system. The asymptotic analytic expression of queue length with FARIMA input is obtained with the appointed buffer overflow probability. It is showed that queue length distribution is influenced by the short-range dependence with small buffer, but the queue length will be asymptotic weibull distribution with increasing buffer size, this is irrespective with short-range dependence. Monte-carlo simulation confirms the validity and efficiency of these results.

Key words: Queue length distribution, Overflow probability, Fractional auto-regressive integrated moving-average(FARIMA), Long range dependence