Chinese Journal of Computers   Full Text
Title	A Kind of P2P Fuzzy Recognition Method Based on Flow Characteristic Description
Authors	SUN Zhi-Xin1)，3) GONG Jing2)
Address	1)(College of Computer of Nanjing University of Posts and Telecommunications, Nanjing 210003) 2)(College of Applied Mathematics & Physics, Nanjing University of Posts and Telecommunications, Nanjing 210003) 3)(State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210093)
Year	2008
Issue	No.7(1252—1260)
Abstract &   Background	Abstract Some characteristics of network application flow are inconspicuous, and they can not be simply identified by keywords, ports+IP. The paper applies fuzzy mathematics theory to P2P traffic identification, and presents a kind of Fuzzy Recognition Method based on Flow Characteristic Description (FCD). FCD fuzzy recognition method analyzes captured network data first, and then describes uniformly key processes. Subsequently, the paper analyzes distributional state of those data in key process, and obtains their membership functions, which can be used as comment aggregates. Finally, fuzzy judgment method will judge whether the data are some kind of network application. The proposed recognition method is put to test through a popular network game. The experiment results verify that FCD fuzzy recognition method can identify network game flow and its accuracy is high. Keywords peer to peer; data flow; fuzzy judgment; flow characteristic; network game Background At present, common P2P flow identification ways include the methods based on detecting default service ports, transport layer, payload, network dynamic characteristic, etc. However, these methods can’t analyze and recognize P2P applications which have inconspicuous characters, or unopened protocols. This paper applies the theory of fuzzy mathematics to P2P flow identification, and proposes a fuzzy identification method based on flow characteristic description (Flow Characteristic Description, FCD). FCD fuzzy identification method first analyzes captured P2P flows, and then describes key flows uniformly. Subsequently, the paper analyzes distributional situation of data in these key flows, and obtains their membership functions, which can be used as comment aggregates. Finally, fuzzy judgment method determine whether the data are some kinds of network applications. FCD fuzzy identification can more accurately recognize P2P applications with ambiguous characters, or unopened protocols, such as some large-scale network games. The research group has many publications of high-quality and patents in this domain.