Abstract:

In this paper, a training method of Multidimensional Support Vector Machines (MSVM) is proposed, and a P2P network traffic identification model is constructed based on MSVM. The model uses MSVM classification to identify all P2P traffic. Data capture module, feature extraction module, data preprocessing module and MSVM training modules are used to classify all kinds of network traffic into P2P traffic and NonP2P traffic. Then by construction of MSVM library, the specific P2P traffic and unknown P2P traffic can be identified. The feature information of unknown P2P traffic is extracted by data acquisition module, feature extraction module, data preprocessing module and MSVM training modules, and then is added to the MSVM support vector library for future P2P traffic identification. Theoretical analysis and numerical experiments show that the constructed model can achieve better identification results with desired accuracy.

Key words: computer systems organization, P2P, multi support vector machine(MSVM), optimal hyperplane