Application Of The Improved PrefixSpan Algorithm In Intrusion Detection

The security of network information system is increasingly significant with the rapid development of Internet technology and wide application of computers. To ensure the security of computer systems and data communications between them, monitor the network intrusion, the researchers have designed various types of safety and security strategy. Network intrusion detection is considered an important aspect of network security, and it have been the research focus in recent years. A higher efficiency requirements for the algorithm which involved is needed, since the dynamic network system security monitoring, mass of network data and the real time requirements. With the development of data mining technology, the mining capabilities has improved greatly with the development of data mining. It has great advantages to be applied to intrusion detection with mining capabilities greatly improve, compared with other technology.The PrefixSpan algorithm, which is broadly applied to data mining field, is one of the most high-efficiency classical algorithms. Taking account of insufficiency of PrefixSpan algorithm, the thesis trys to optimize the algorithm by reducing frequency of exchanging between the memory and the external memory in the Prefix part, and reducing the size of the projection database by bi-level projection and discarding the non-frequent items which created in the process of sequence patterns mining. The result of test demonstrates that the operating efficiency is enhanced more than 30%. The conclusion of the experimental analysis shows that the improved algorithm is applicable to the invasion detection.