Improving Wu-Manber: A Multi-pattern Matching Algorithm

The likelihood of shifting text sliding window based on Boyer-Moore SHIFT table at each comparison point heavily affect the efficiency of the Wu-Manber algorithm. This paper introduced a second Boyer-Moore SHIFT table into regular Wu-Manber algorithm in an effort to increase the likelihood of shifting text sliding window at each comparison entrance; so the costly exact string comparison is avoided most of the time, leading to a faster multi-pattern matching algorithm. Our experiment result shows that, the improved Wu-Manber algorithm is 50% to 80% faster than regular Wu-Manber algorithm, making it suitable for gateway-based online content scanning scenarios with medium number of signature patterns.