多重分形网络流量判别方法和产生方法  

Ways of indentfying and producing multifractal networks traffic

作　　者：张博[1] 陈魏鑫[1] 俞柏锋[1] 

机构地区：[1]国家数字交换系统工程技术研究中心,河南郑州450002

出　　处：《计算机工程与设计》2009年第3期566-568,618,共4页Computer Engineering and Design

基　　金：国家863高技术研究发展计划基金项目(2004AA103130)

摘　　要：通过对网络流量多重分形特性的研究,提出了粗粒化、H?lder指数、奇异谱3种判别分形网络流量的方法,和RMD叠加、基于树型2种人工多重分形序列的产生方法。采用真实网络流量数据对所提出的3种判别方法进行检验,结果表明3种判别方法均可有效的判别网络流量的单重分形和多重分形特性。然后用该方法检验2种人工序列,检验结果人工序列均具有多重分形的特性,可以拟合真实网络流量。Third ways ofindentfying fractal networks traffic are proposed including coarse grain methed, Hoelder exponent method, and singular spectrum method, and two ways of producing got-up multifractal list is presented including RMD degistration methods and treebased method via investigating networks traffic multifractal characteristics. Then real networks traffic data is used to test this third ways ofindentfying fractal networks traffic. The testing result show these ways can indentfying networks traffic single-fractal and multifractal characteristics. Then indentfying ways is used to test got-up list. The result of testing is two got-up lists have multifractal characteristic and could fit real networks traffic.

关 键 词：网络流量 人工序列 多重分形 单重分形 拟合 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]