Improvement Of Deep Packet Service Identification In P2P Network

In recent years, the use of P2P(Peer-to-Peer) technology is becoming more and more widespread. However, P2P brings convenience to people, but also caused many problems. For example,the P2P network characteristic, e.g."the bandwidth gobble up", results in the huge depletion of bandwidth, even causes the network congestion, lowers the network performance, and makes the network service quantity inferior. Therefore, P2P service identification is significant.This paper briefly introduces the common method of service identification, port identification method, machine learning identification method, analysis their working principle,comparing their advantages and disadvantages, then focus on DPI (Deep Packet Identification) method,which has the highest accuracy in service identification. Through research, there is two problems in DPI method: 1) over-reliance on services application layer signature, but it is too difficult to get the services application layer signature. 2) AC(Aho-Corasick) automaton algorithm is an efficient multi-pattern matching algorithm, often used in the pattern matching of DPI method. However, due to the principle of AC automaton algorithm is to construct a matching template offline before pattern matching, but when found a new application layer signature, we must use all the application layer signature matches to re-construct a template, so poor scalability of AC automaton algorithms reduces its efficiency in the application of DPI method.For these two problems, the author proposes an algorithm based on suffix array to get the service application layer signature efficiently, while designs an incremental construction matching template of AC automaton algorithm which improves the scalability of AC automaton algorithm in DPI method .The paper proves the correctness and efficiency of the two algorithms from a theoretical point.Then the author designs a simulation system of service identification using DPI method,the two algorithms are used in the system,and proves the correct and efficient from a practical point.