Analysis And Identification Of P2P Streaming Media

P2P (Peer to Peer) is one of the most popular Internet technologies in recent years, which is widely applied and rapidly developed in the fields such as VoIP, flie-download, coordinately computing. Also in the area of streaming media, the technology also reflects the strong growth momentum, not only the exclusive software such as PPLIVE, PPS, UUSee, QQLIVE, but also some browse-based websites such as Youku and Tudou are all the specific applications. These applications and Web sites provide users with various forms of entertainment, such as the popular TV drama, exciting live sporting events and other TV programs. While P2P streaming media facilitating users, it occupies large network bandwidth volumes, which causes severe security and quality of service problem, and puts the network operators in the dilemma of repeatedly "congetion-construction-congesion" circle. For better use of the Internet, the traffic characteristics analysis and traffic identification of P2P streaming media has become a hot research direction of many studies. The current main identification methods contain the recognition technology based on the probability characteristicsc and the one based on the deterministic characteristicsc. The thesis presents the research on timely identification for P2P streaming media traffic, including the following topics:Research based on the probability characteristicsc for timely P2P streaming media traffic identification:Traffic identification based on the probability characteristics is the current hot research direction, many researchers have combined with machine learning to identify P2P traffic. However, most of these methods identify the application types after the flow has ended. So these methods are mainly used for off-line network traffic analysis, network optimization, and user behavior analysis. Thus we analyze the characteristics of P2P streaming media flow, propose a real-timely indentification approach through the sub-flow characteristics. This approach ignores the identifation of the non-media data flows, soloving the low recognition accuracy problem of the identification based on the probabilistic characteristics. At the same time, tha approach uses the sub-flow which only contains parts of the whole flow packets for identification, making the identification before the end of the flow.Research based on the deterministic characteristicsc for timely P2P streaming media traffic identification:Traffic identification based on deterministic characteristics is one of the most widely used identification method with higher accuracy. The method identifies the application types mainly by the signatures in the packets. But the signatures are mainly received by artificial protocol and packets specification, due to the extensive use of private protocols, the signatures are more and more difficult to summarize. In this paper, we analyse and implement the Automatically Generating Signatures for Applications algorithms, and use it to comprehensively abstract the signatures of the four popular P2P streaming media applications, and analyse signature positions to study the real-time effect of methods.Timely two-strategic traffic identification for P2P streaming media:After the completion of the above two real-time traffic identification methods for P2P streaming media. A real-time two-strategic identification is proposed. This method combines the advantages of the two identification methods, improving the recognition accuracy further, and ensuring the highly accurate identification for the latest version of P2P streaming media software.