A Useful Anomaly Intrusion Detection Method Using Variable-length Patterns and Average Hamming Distance

Abstract

Intrusion detection techniques at the level of system processes are discussed, and a new method named V-AHD (Variable-length Average Hamming Distance) is presented, which can be used to monitor and calculate deviation to discriminate between normal and abnormal sequences of system calls. For the reason that fixed-length patterns can not describe the system behavior correctly and its inability to represent long meaningful substrings, the V-AHD method use Teiresias to get variable-length patterns and construct the normal set. Then the algorithm for detection is described in detail, and the pseudocode is also given. The method has some advantages, such as algorithm simplicity, low overhead of time, high accuracy and real-time detection. The prototype experiments with four attacks prove the validation of the method, which has high True Positive Rate and low False Positive Rate.