基于瀑布过程的网络流量模型

Network Traffic Model Based on the Cascade Process

分形或自相似模型可以很好地描述网络流量过程的长程依赖性 ,却无法真实地刻画网络流量过程在小尺度上的奇异性 .论文从另一个角度 ,研究并建立了基于瀑布过程的瀑布模型 ,并对真实的网络流量数据进行模拟分析 ,发现瀑布模型从一定程度上解释了实际网络流量中的尺度特性 ,它能准确地刻画网络流量在小尺度上的奇异性 .通过对模拟数据的尺度分析证明 。

Fractal or self-similar models can characterize the Long-Range Dependence of the network traffic. However, they can not capture the singularity in small scale. This paper adopts a different way and build the multifractal model based on the cascade process: the cascade model. This paper also analyzes the real traffic and finds that the cascade model can explain the scaling feature of the real traffic and accurately capture the singularity of network traffic in small scale. The scaling analysis of simulated traffic also proves that the cascade model possesses the capability of characterizing multifractal feature.