The P2P Applications Analysis And Control Based On Flow Characteristics

With the rapid development of the internet, kinds of the P2P applications emerge. Recently, P2P is widely applied to file-sharing, streaming media, instant messaging and so on as a new network application model. The increasing of the P2P applications causes the huge consumption of the network bandwidth, and also causes the flooding of the virus on the Internet, which will bring great inconvenience to users. So achieving an accurate identification of P2P traffic and reasonable control is of great significance for network supervision.In this thesis, we firstly analyze the current situation of P2P technology and introduce its basic theory, and then sum up the existing P2P traffic identification method. In addition, by analyzing the realization mechanisms and message formats of the three most widely used software protocols namely BitTorrent, PPlive and eMule, we extract the feature strings of the three protocols and use them to carry through the identification. At the same time by comparing the data connection feature of the three protocols to that of the HTTP protocol, we summarize the common characteristics of the data connection. Finally, we put forward an idea for the recognition which gives first place to feature strings and second place to the data flow link feature recognition. And we introduce the design and the implementation of the data stream recognition system (mainly for the encrypted data stream) in detail, mainly by data stream capture, flow recognition and traffic control. The test proves that this system can identify P2P encrypted data stream better.