Transport Layer Identification Of P2P Traffic

P2P traffic is on an enormous role in promoting the progress of the Internet, while also consuming a lot of network resources, impeding the normal operation of network services. It is necessary to identify P2P traffic and control. Although the P2P traffic identification method based on application layer characteristics has been widely used in a variety of commercial traffic monitoring system, but it still has unavoidable defects. The new P2P applications can not be identified. the more and more P2P applications use encrypted payload in future. these methods also are not able to identify. Therefore, the transport layer identification of P2P traffic will be of great significance. This paper focuses on two kinds transport layer identification of P2P traffic--based on connection pattern identification method and traffic pattern method identification.For the connection pattern identification method, first, the methods mechanism and implementation steps are researched; Second, the extended method based on the connection pattern is proposed by summarizing domestic and foreign latest technology; Finally, the experimental results show that the connection pattern method has a high accuracy, but algorithm needs to traverse the flow table, it applies only to off-line analysis.For the traffic pattern identification method, first, the data mining technology in traffic identification is researched; Second, K-means algorithm is more suitable for traffic identification technology by comparing three clustering algorithms; Third, the parallel strategy of K-means algorithm is proposed; Finally, the parallel K-means algorithm is achieved by using OpenMP technology. The experimental results show that, in the dual-core environment, the program can improve the operating efficiency of nearly 30%.