Research On Peer To Peer Traffic Identification Method Based On Artificial Bee Colony Algorithm And Wavelet Support Vector Machine

In recent years, with the rapid development of the network, the peer-to-peer(P2P) network has become one of the most important technologies in the Internert. P2 P application has brought the development and prosperity of network,but it also produces many security problems to the network management, such as network bandwidth congestion problems, the spread of illegal content etc. Hence, detecting and controlling P2 P traffic has attached much attention to academic community and network operators.P2P traffic identification is a classification problem for 2 classes in nature, the selecting of traffic characteristics and the method of classification have a great influence on the accuracy of P2 P traffic identification, in order to improve the accuracy of identification, the paper mainly incudes the following work:(1) The feature selection of P2 P traffic based on artifical bee colony algorithm. The accuracy of P2 P traffic identification is usually low when use a single feature. So it is very necessary to use more features to improve the accuracy of P2 P traffic identification. But too many features of traffic identification will bring the problem of curse of dimensionality. This is not only can not improve the efficiency of traffic identification, but also collection of excessive traffic features will increased the workload, the real-time of traffic identification is difficult to become. Therefore, the paper proposes the artificial bee colony algorithm for feature selection; it can get the best characteristics classification performance.(2) The P2 P traffic identification based on wavelet SVM and artificial bee colony. The support vector machine can effectively avoid the phenomenon of over-learning in machine learning, but the SVM with traditional kernel only on a single scale to classify the sample data. The wavelet kernel function can approximate any function with high accuracy and is conductive to deal with multi-scale to improve the generalization of SVM. So the wavelet functions can be introduced to contruct the kernel function of SVM and used for P2 P traffic identification. At the same time the parameters of SVM(penalty factor C and kernel parameter) have significant impact on the classification accuracy of results, so the selecting of parameter optimization method is very important. The common intelligent optimization approaches such as genetic algorithm, particle swarm optimization algorithm are easy to fall into the local optimum. The global optimization ability of the artificial bee colony algorithm is very good, so the paper use the algorithm to optimize the parameters of support vector machine.Finally, the proposed feature selection and SVM parameters optimization method are tested on UCI database and real campus network of P2 P data to verify the effectiveness of the proposed method. The results are compared with the existing GA algorithm and particle swarm optimization algorithm. The combination of wavelet kernel function with SVM classification model can be used for P2 P traffic identification. The final results shows that the use of artificial bee colony algorithm with feature selection can get the best feature classification result, the wavelet SVM has better result of the P2 P traffic identification.