中文摘要：

提出了一种基于主机行为关联的加密P2P流量实时分类方法，该方法基于P2P系统中节点间的连接关系，以一定的先验知识为初始条件进行节点发现，并根据网络行为不断进行迭代检测，持续发现P2P网络中的新节点及其对应的流量，从而达到对加密P2P流量实时分类的目的。真实流量环境中的对比实验表明，该方法对典型加密P2P流量的分类准确率、召回率均超过95％，计算代价小、性能高，不依赖于内容检测，不侵犯用户隐私，能有效应用于实时流分类环境中。

英文摘要：

A real-time method for classification of encrypted P2P traffic based on host behavior association is proposed. Initialized with some priori knowledge, this method finds the nodes in a P2P network according to the nodes＇ con- nection relationship, and constantly finds the P2P network＇ s new nodes and their corresponding traffic by examining the iterating over nodes＇ network behavior to achieve the real-time classification of encrypted P2P traffic. The re- suits of the experiment on a real campus network showed that, besides low computational cost, both the accuracy and recall rate of the proposed method were above 95 %. Meanwhile, by classifying traffic without content inspection, the proposal violates no users＇ privacy, so it can be used flexibly in high-speed network environments.