一种基于SVM的P2P流量检测方法研究  

Research on a Method of P2P Traffic Detection Based on SVM

作　　者：姜伟[1] 王春枝[1] 罗宏芳[1] 陈宏伟[1] 

机构地区：[1]湖北工业大学计算机学院,湖北武汉430068

出　　处：《湖北工业大学学报》2010年第2期48-50,共3页Journal of Hubei University of Technology

基　　金：湖北省自然基金项目(2009CDB100);武汉市晨光计划项目(201050231084)

摘　　要：提出了一种基于SVM的P2P流量识别模型,选取P2P流量的上/下行流量比,平均流速率、流持续时间、传输字节数、平均数据包长度等5个特征,对网络流量的P2P流量进行识别.实验结果表明,该方法能够有效地检测网络流量中的P2P流量,准确率达到96%.As a machine learning algorithm that has developed in recent years,SVM can be used to solve problems of highly nonlinear classification and regression.The paper proposes a P2P traffic identification model based on SVM.It identifies P2P traffic in network by choosing five characters of P2P traffic such as traffic ratio of up-going and down-going,average rate of traffic,traffic lasting hour,number of transmission bytes,and average height of data packet.Results of the experiment show the method can detect P2P traffic in network effectively with an accuracy of 96%.

关 键 词：流量特征 支持向量机 对等网络 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]