Research And Implementation Of P2p Traffic Identification Methods

With the unique advantage of sharing resources, the P2P technology has beenwidely applied to the fields of network TV and resource downloading. A variety of P2Papplications enrich people’s network life, but P2P applications seriously seize thebandwidth resources. According to statistics, P2P traffic has occupied60%-80%of thetotal Internet traffic, which poses a serious threat to the operation of network. Thereforepropose an efficient and accurate method to identify and detect P2P traffic makes greatsense.This thesis focuses on the study of P2P traffic identification. It firstly reviews theprinciples of P2P technology and current status of P2P traffic identification methods,and then deeply studies the DPI technology and the traffic feature detection technology.During the study of DPI technology, the thesis summarizes the principles of designingsignatures and the methods of designing signatures; according to the above principlesand methods, it designs the signatures for4widely used protocols. During the study oftraffic feature detection technology, the thesis proposes a P2P traffic identificationmethod based on the characteristics of udp session and design the schema of identifyingP2P traffic by classifying tcp and upd flow.Based on the deep study and analysis of P2P traffic identification methods, thethesis proposes a comprehensive p2p traffic identification model, which combinesseveral kinds of traffic identification methods to detect P2P traffic. Different methodextracts different characteristics from network flow to identify P2P traffic, and methodscomplement each other to achieve high accuracy.The model can not only accuratelyidentify variety of widely used P2P protocols but also has the ability to detect unknownand encrypted P2P traffic, which can fully meet the needs of P2P traffic identification.And it describes the design and implementation of the model in detail, which includesthe design principles of model, the deployment environment, the overall structure ofmodel, the key technologies and functional modules.To evaluate the feasibility and effectiveness of the P2P traffic identification model,we design and implement the prototype system. The system is deployed in a small LAN environment to test its performance and effectiveness. The experimental results showthat this model can accurately and efficiently identify P2P traffic, which commendablyimplement the purpose of the P2P traffic identification and control.