基于对数无限可分级串框架Web网络流量分析

Web Traffic Analysis Based on Log-Infinitely Divisible Cascades

由于Web流量占据了网络流量的主要成分,它的尺度特性必然对整体流量特性产生一定的影响。而对数无限可分级串框架能描述数据流全范围下的多尺度行为,因此可在该框架下分析Web流量与TCP流量的多尺度行为。通过几个对比实验,发现Web流量并不一定就决定了TCP流量的尺度特性,异质的非Web流量的尺度分段交接点大于同质的Web流量的交接点。而且在相近的时间内,Web流量的尺度特性不受一些异常流量的影响,是相对稳定的。

Web traffic,which predominates in network traffic,should have effects on the scaling properties of whole traffic.The log-infinitely divisible cascades method in scale analysis of turbulence could describe the multi-scaling behavior of data in full range,which we used to study the scaling properties of web traffic and TCP traffic.Based on several designed experiments,it is leant that web traffic could not definitely decide the scaling properties of TCP traffic,and the scaling joint of non-web traffic which is heterogeneous in protocols is larger than that of homogeneous web traffic.The scaling properties of web traffic are relatively steady,which could not be influenced by abnormal traffics.