Research on a Method of P2P Traffic Detection Based on SVM

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%.