Internet拓扑抽样探测的偏向性研究

Bias research on sample exploration of Internet topology

利用traceroute探测网络拓扑结构是研究Internet拓扑属性的主要方法,然而由于这种方法的局限性,使得该方法对正确分析Internet拓扑属性产生很大影响。采用几种网络模型来分析traceroute探测得到的抽样网络属性的偏向性问题,包括节点发现率、边发现率和度分布等。同时针对网络探测参数设置的局限性,还定性分析了在总体抽样粒度不变的情况下,如何设定探测源和目标节点的密度才能达到优化探测网络特征属性的目的。

Traceroute exploration is the main method in Internet topology discovery.However,due to the localization of traceroute,the results of Internet topology properties are biased.This paper introduces several network models to analyze the bias problem on network properties sampled by traceroute exploration,including node discovery proportion,edge discovery proportion and node degree distributions.Based on the limit of setting network exploration parameters,it also discusses the optimization problem of network exploration on deploying the sources and destination density from a quality view.