微信数据流量特性的全面分析

Analysis of WeChat Data Flow Characteristics

本文以微信应用为背景,通过识别分类提取微信流量,分析微信业务的流量特性。研究表明,微信流量具有高可变性和自相似性,无法用传统的泊松分布来统计。此外,微信流量还具有明显的突发性,呈现幂律特性,可以通过重尾分布来统计。本文对比不同的重尾分布,发现pareto分布更能准确刻画微信流量的重尾特性。

In the context of WeChat application,the WeChat traffic is extracted by identifying and classifying,and the traffic characteristics of WeChat service are analyzed.The research has shown that the WeChat traffic has high covariance and self-similarity and cannot be characterized by the traditional poisson distribution.In addition,WeChat traffic also has obvious burstiness,showing power-law characteristics,which can be characterized by heavy-tailed distribution.The different heavy-tailed distributions are compared and finds that the pareto distribution can more accurately describe the heavy-tailed characteristics of WeChat traffic.