摘要： 利用从实际网络获得的数据，提出基于流的P2P网络特性模型。采用图形方法和概述统计识别样本所服从的分布族，使用可视化图形方法和假设检验方法对统计分布模型进行拟合优度检验。分析结果表明，流持续时间的分布模型可以用对数正态分布精确表示，混合对数正态分布可以有效拟合流长和流传输速率分布的分布模型，且P2P应用的流长和流持续时间没有高度相关的关系。

关键词: P2P应用, 网络流量, 流, 对数正态分布, 混合分布

Abstract: This paper proposes a P2P network characteristic model based on flow obtained from the actual network. It uses diagram and summary statistics process to recognize the distribution of sample, designs the statistical distribution model, and uses visual graphic method and the hypothesis testing method to do goodness of fit test for statistical distribution model. Analysis results show that the lognormal distribution can accurately express the flow duration, while flow size and flow transmission rate can be reasonably modeled by lognormal mixture distribution, however P2P application flow size and flow duration are not highly related to each other.

Key words: P2P application, network traffic, flow, lognormal distribution, mixture distribution

中图分类号: 

• N945.12