Improvement on Anomaly Intrusion Detection Algorithm Based on V-Gram
Jifeng Yu, Jingli Zhou · Physics Procedia · 2012
Short sequences of system calls of running processes are good data sets for anomaly detection system. This paper analyzed the insufficiency of fixed-length patterns (N-Gram) and variable-length patterns (V-Gram) in processing sequences of system calls. A new model of Multi Wildcards V-Gram (MWV-Gram) with redundancy controlling mechanism is presented. Experimentation results indicate that the pattern database is reduced and detection efficiency is enhanced by the improved algorithm.