基于瀑布过程的多重分形模型特性分析与检验

Multifractal Model Characteristic Analysis and Test Based on Cascades Process

通过连乘瀑布过程的构建方法,采用对称的Beta分布拟合瀑布过程的乘数因子以及独立的pareto分布拟合瀑布过程的随机变量,建立网络流量多重分形模型。求解该模型合成流量的均值、方差、协方差的表达式,得出其具有正值、长相关、非二阶平稳特性。采用LBNL采集的数据DEC-PKT-1对其进行拟合检验,证明该模型在拟合真实网络流量时的有效性。

This paper constitutes net traffic joint multifractal model based on multiplicative cascades process via symmetrical Beta distribution fitting multiplicator factor and absolute pareto distribution fitting random variable. It settles average value, variance and covariance of mixture traffic, and analyzes the model＇s pure, long-range dependence and not two rank stableness characteristic, and uses DEC-PKT-1 of LBNL data to fit and test. Result proves this model is operative on fitting real networks traffic.