Research Of P2P Traffic Identification Based On DPI

With the rapid development of P2P (peer-to-peer) network applications technology, it becomes a heavyweight of Internet applications. In one side, it brings convenience to Internet users, but in the other side, it also seizes the majority of the network bandwidth for a long time and devours a large number of network resources. It makes the network congestion and increases the burden on the network. Eventually, it weakens the ability of network applications for all users. Therefore, the efficient and rational control for P2P network is an important method to improve the network environment. Among them, the identification of P2P network data is a prerequisite for the control. This paper aims at finding out ways to improve the efficiency of DPI testing.In order to detect and control the P2P traffic efficiently, the DPI deep packet scanning technology is studied deeply and more detailed study spend on the current popular string matching algorithm for multi-mode:AC (Aho-Corasick) algorithm, Wu-Manber algorithm and SBOM algorithm in this paper. It analyses the principle of the three algorithms. Three algorithms are used in DPI scanning under the same testing environment and the pretreatment time, analysis speed, degree of accuracy from this testing were compared. Based on these experimental data, the superiority-inferiority of these three algorithms in different conditions is discussed and their relative scopes of application are summarized. But DPI identification technology is difficult to apply in high-speed networks due to its slow execution speed. To solve this problem, this paper suggests six sampling strategies:invariable random sampling, invariable mechanical sampling, time random sampling, time mechanical sampling, speed random sampling mode and speed mechanical sampling. Discussing and testing how these strategies affect the DPI identification system in different network environments. Test conclusions shows that six sampling modes reduce the burden on the system in different degrees and maintain a high identification ratio of P2P data stream. Then trust mechanism based on sampling method is proposed to reduce the burden on the hardware and network latency, so that less loss and higher efficiency operation are premised on the accuracy the DPI. We achieve these following results:1. The comparative study of three algorithms AC (Aho-Corasick), Wu-Manber and SBOM was accomplished. The superiority-inferiority of these three algorithms in different conditions is discussed and their relative scopes of application are summarized.2. Slicing mechanism sampling method is designed to explore the mathematical theory of this method and the software design process. 3. This article achieved the sampling method based on slicing mechanism and on this basis fully proved the feasibility and effectiveness of sampling methods by a large number of actual data testing. It concluded that the different methods demonstrate differences in different network environment.4. The new DPI sampling method based on trust mechanism is proposed. it is a further derivative of slicing sampling methods. In theory, new types of sampling methods is described in detail, including how to build a new sampling method steps, how to achieve the trust mechanism, how to set up the trust rules and how to design change rules of sampling ratio of. This design has the nature of innovation and invention.