A P2P traffic identification approach based on the optimal support vector machine and genetic algorithm

摘要

Peer to peer (P2P) traffic identification is a hot topic in the P2P traffic management. P2P traffic identification method based on support vector machine (SVM) is one of the most commonly used methods. However, the performance of SVM is mainly affected by the parameters and the features used. The traditional method is to separate the SVM parameter optimization and feature selection problem, it is difficult to obtain the overall performance of the SVM classifier. Thus, an approach of P2P traffic identification based on the optimal Support Vector Machine and Genetic Algorithm is put forward. It takes the parameters of SVM and the feature selection problem can be treated as the simultaneous processing of the optimization problem and the optimal parameters and feature subset of the whole performance can be obtained. The proposed approach is validated on P2P data. The results show that the approach has very good classification accuracy, and it can effectively detect the P2P traffic in the network traffic on the basis of obtaining the optimal parameters and feature subset.