A Peer To Peer Traffic Identification Approach Based On Bacterial Foraging Algorithm And Wavelet Support Vector Machine

P2P technology has been widely used and extensively developed, since internetitself is based on point to point transmission. But it seriously affects user managementand normal network operation of network operators, with rapid increase of a variety ofP2P applications. The serious problems of P2P applications, such as low bandwidthutilization, network congestion, intellectual property issues, resource management,safety problems, and mutation and uncertainty of P2P network, which make theidentification of P2P traffic becomes more and more difficult. So it has been of greattheoretical and practical value to identify and control P2P traffic effectively, accuratelyand fast. As Support Vector Machine (referred to as SVM) has special advantages ofavoiding local optimum and overcoming dimension disaster, this paper proposed anovel model based on bacterial foraging algorithm and wavelet support vector machinefor P2P traffic identification, with basis of current P2P traffic identification technologyanalysis and summary. The final experimental results and comparison analysis provedits feasibility and effectiveness. The main research contents and contributions of thisthesis are as follows:Firstly, we introduced bacterial foraging algorithm to optimize the two SVMparameters, which can improve the identification rate of SVM and real-time capability.Compared with existing genetic algorithm and particle swarm optimization algorithm,the experimental results demonstrate that the optimized SVM can significantly improvethe performance of traffic identification.Secondly, wavelet analysis can be a good solution to settle the mutation anduncertainty of P2P network, because of the superiority in aspects of local analysis andmutant signal processing. Thus we can choose a suitable wavelet kernel function toimprove the performance of SVM. Through comparative experiments of a variety ofcommonly used core functions and wavelet kernel function, we can prove that thewavelet kernel function for P2P traffic identification based on bacterial foragingalgorithm and SVM has a higher identification accuracy and stability.