Research And Design Of P2P Traffic Control In Firewall

With the extensive use of P2P applications, it has occupied a large amount ofnetwork bandwidth. P2P traffic control has become an indispensable component of thefirewall,inorder tobe moreeffectivelyontheprotectionandmanagement ofnetworks.The identification of P2P traffic is the prerequisite and basis for management. Atpresent, a lot of firewall productions identify P2P traffic by analyzing port numbers orpayload.However,withtherapiddevelopmentofP2Ptechnology,theP2Psoftwarecannow penetrating the firewalls by using dynamic ports, and protocol encryptiontechnology.Based on the background discussed above, this paper presents a scheme of P2Ptraffic control, which is an integrated scheme of machine learning approach andpayload-based analysis. The use of machine learning approach is to identifyP2P trafficand manage network bandwidth, while the use of payload-based analysis is to identifyandmanagethecorrespondingflowbehavior.The first two chapters describe the development of P2P traffic control, and pointout the problems on P2P traffic identification and management. Then, in order to solvethe problems, this paper proposed a composite of P2P traffic control scheme in chapterthree. In chapter four, the paper chooses a representative data as study data, usesK-Means algorithm to classify the study data and establish the model, and thenidentifies P2P traffic by NN methods. The fifth chapter describes one or morecharacteristic values of five typical P2P protocols, and proposes the payload-basedtraffic control module by using a fast string searching algorithm. The last chapteranalyses the P2P control scheme described in this paper, explains its advantages anddisadvantages,andbringsforwardthenewresearchdirectionsinthefuture.