The Study Of P2P Traffic Identification Methods

applications have experienced a pretty rapid development and occupied many network application fields. P2P applications, especially filesharing and multimedia, adopt strategies such as using dynamic ports and encrypting payload to escape from traditional traffic identification methods, causing many difficulties for traffic management and analysis.Based on the study of many other P2P traffic identification methods, this thesis proposes an identification method based on twofold features, namely traffic features and payload features. This method has proper accuracy and promising efficiency. Moreover, by importing Support Vector Machines to P2P traffic identification fields, this thesis proposes two methods called Iterative Method and Smooth Method. Iterative Method improves the identification accuracy and compresses the training data by means of iterative training; Smooth Method reduces the problem complexity and improves the accuracy using smooth processing. Experimental results show that the two methods both have the ability of accurately identifying P2P from unknown network traffic. Finally, based on the algorithms which have been optimized for large tasks, a method which can classify P2P in application level is proposed. Experimental results show that this method not only has high efficiency, fitting for realtime identification, but also achieves high accuracy by carefully tuning parameters.