网络业务量的一种稳定多分形模型及其检验

目前网络业务量研究中有自相似和多分形两种主要的先验模型。自相似模型简洁，运用单一的H 指数来描述业务量特性，但它并不能完全表达业务量的信息；多分形模型更准确，但它运用函数形式的多分形谱来刻画信号，不便于实际分析。本文在无穷可分层叠的理论框架下提出了网络业务量的一种稳定多分形模型。运用小波变换方法，通过多个业务量记录的检验，证实此模型从尺度行为和边缘分布两方面都能准确刻画业务量。作为一般多分形过程的特例，该模型把多分形指数的表达式确定下来，并归结为4个参数的估计。

A Stable Multifractal Model of Network Traffic and Its Verification

There have been two a priori models up till now concerning network traffic: self-similar and multifractal models. The self-similar model is simple enough to characterize network traffic with the single exponent H , however it can not thoroughly describe the contents. The multifractal model, which is more accurate, is not practical because it uses multifractal spectra as functions to represent signals. Therefore, on the basis of the framework of IDC theory, this paper proposes a stable multifractal model of network traffic. It is tested on several traces by using the method of wavelet transforms that the model fits traffic precisely in terms of both scale behaviors and marginal distributions. As a specific case of multifractal processes, this model, by estimations of four parameters, can specify the representation of the multifractal exponent.