Intrusion detection based on cross-correlation of system call sequences

A new light-weight approach, based on the cross-correlation of system call sequences, is presented to identify normal or intrusive program behavior. The program behavior is represented by the cross-correlation value which can be used to indicate the similarity between two sequences. If two sequences are same, the cross-correlation between them will achieve the maximum value. This method of characterizing program behavior by using cross-correlation offers significant computational advantages over HMM (hidden Markov model) or NN (neural network) methods due to the absence of unnecessary training process. Our experiments using UNM (University of New Mexico) audit data show that the cross-correlation based method can effectively detect intrusive attacks and achieve a low false positive rate