Abstract: A novel frequent item mining algorithm based on multi-dimentsional clustering is proposed, which uses clustering theory to mine significant flow and its rules automatically in network. On basis of this, P2P suspected identification is conducted to these flows. Combined with application layer feature recognition technology, highly suspected P2P significant flow is filtered, and unknown P2P traffic detection is implemented. Experimental results show this method is effective.

Key words: Peer-to-Peer(P2P) network, frequent item mining, traffic identification

摘要： 提出一种基于多维聚类的频繁项挖掘算法，利用聚类思想自动挖掘网络中的显著流量及其规则，并在此基础上，对显著流量进行P2P疑似性判别，同时结合应用层特征识别技术，对高度疑似的P2P显著流量类进行过滤，实现未知P2P流量检测。实验结果表明，该方法是有效的。

关键词: 对等网, 频繁项挖掘, 流量识别

CLC Number: 

• TP393.09