自相似网络流量差分分析新方法  

New Analysis Method for Self-similar Network Traffic Based on the Differential Analysis

作　　者：李立[1] 喻莉[1] 朱光喜[1] 白云[1] 

机构地区：[1]华中科技大学电子与信息工程系/武汉光电国家实验室,武汉430074

出　　处：《计算机科学》2008年第4期112-114,共3页Computer Science

基　　金：国家自然科学基金资助项目(60502023)

摘　　要：多媒体网络流量具有重尾特性及自相似特性,已得到了广泛的认同。而Alpha-stable过程作为Gaussian过程的推广能很好地描述重尾特性及自相似特性。在简要介绍了Alpha-stable分布和Alpha-stable过程的基础上,本文研究分析了当前基于线形分形稳定噪声(LFSN)理论的网络业务流量模型,并提出了基于差分分析的新方法及基于差分分析的LFSN流量模型参数估计新方法。对Bellcore实验室采集数据的拟合验证结果表明,该估计方法能准确确定模型参数。The fact that multi-media network traffic is with heavy-tail character and self-similar character has been known to researchers. At the same time, Alpha-stable process, the extension of Gaussian process, has the advantage of modeling heavy-tail character and self-similar character. Alpha-stable distribution and Alpha-stable process are introduced at the beginning of this paper. Then after analyzing of current network traffic models based on Alpha-stable process and Linear Fractional Stable Noise （LFSN）, we advance a differential analysis method and a new parameter estimation method based on difference analysis. Results of analysis of actual traffic data from Bellcore show the method can get parameters precisely.

关 键 词：网络流量建模 Alpha-stable分布 线形分形稳定噪声 差分分析 参数估计 

分 类 号：TP393[自动化与计算机技术—计算机应用技术] TN918.1[自动化与计算机技术—计算机科学与技术]