Large-scale Communication Network Behavior Analysis and Feature Extraction Using Multiple Motif Pattern Association Rule Mining

Minimize false positive and false negative is one of the difficult problems of network behavior analysis. This paper propose a large-scale communications network behavior feature analysis method using multiple motif pattern association rule mining, analyze multiple behavior feature time series as a whole, produce valid association rules of abnormal network behavior feature, characterize the entire communication network security situation accurately. Experiment with Abilene network data verifies this method. Key-Words: Network behavior analysis, principal components analysis, time frequency analysis, multiple motif pattern association rule mining.