决策树在P2P网络截包分析中的应用

Application of Decision Tree in Analysis of P2P Network Packet-Capturing

为了更高效准确地检测到P2P数据流,将目前在数据挖掘等领域比较成熟的决策树分类技术应用到截包分析研究中.在实际应用中,通过对流的特征属性进行计算统计处理作为训练样本集建立决策树,并对建立的决策树进行剪技优化,实验表明决策树分类技术更能快速准确地定位P2P数据流,在实时处理大量数据方面更能体现决策树分类技术的效率与准确度.

In the area of detection technology for P2P（Peer to Peer） flow,the decision tree（DT） classification techniques which are relatively mature in data mining area is applied to the detection technique of P2P flow in order to detect P2P data flow more efficiently and accurately.In practical applications,dealing with the data flow as a feature attribute is conducted in order to build training sample set and to create a decision tree.Meanwhile,the established decision tree is optimized by pruning tree branches.The experimental results show that the decision tree classification techniques can be located rapidly and accurately to P2P data flow,and its accuracy and efficiency can be reflected more clearly in some aspects of the treatment of large amounts of data than others.