The Research And Application Of Intrusion Detection Based On Sequential Pattern Mining

Recently, with the increasing network flux, the intrusion detection system based on data mining technique has been researched widely. How to integrate effectively data mining algorithms into intrusion detection system is one of the problems to be solved for intrusion detection technique.At present, research on the algorithm in intrusion detection based on data mining technique is mainly focused on four aspects: applying association analysis algorithm to mine rules among different attributes of records, applying sequence analysis algorithm to find out the sequential patterns among different records, applying classification analysis algorithm to forecast whether normal or abnormal new audit records are, and applying clustering algorithm to get the wanted clusters for new network data.The writer aims at the research of sequential pattern mining algorithm that fits into IDS (Intrusion Detection System) and makes the work as following:Analyzing the latest research progress and main problems existed of IDS, researching advantage of data mining technique applied to IDS, and analyzing disadvantage of IDS based on data mining technique.Researching sequential pattern mining algorithm deeply, and improving a sequential pattern mining algorithm.Researching a framework of IDS based on sequential pattern mining, then applying the PrefixSpan algorithm in this model.Finally some experiments has been performed, and the experimental results show the improved PrefixSpan algorithm improve the time and space efficiency and decrease the number of rule and heighten the availability of rule.