A Word-length Based Wu-manber Multi-pattern Matching Algorithm

WANG Yong-jin,GU Nai-jie,REN Kai-xin
DOI: https://doi.org/10.3969/j.issn.1000-1220.2013.07.040
2013-01-01
Abstract:Multi-pattern matching is one of the most important operation in the string processing system,while Wu-Manber algorithm performs best on average.To tackle the problem of performance reduction w hen Wu-Manber algorithm deal w ith short patterns,this paper presents a multi-pattern matching algorithm matched by w ord length.Enhanced algorithm runs on 32-bit machine.By changing match unit from one character to one w ord size,the access time is reduced.Taking advantage of the characters of machine w ord storage,w e speed up the calculation of a block hash,and greatly improve matching performance w ith the existence of short patterns.When the shortest string length is less than 6,the improved algorithm's search time is reduced by 40% compared to the original Wu-Manber algorithm.When the shortest string length is equal to 2 or 3,the search time is reduced by more than 60%.
What problem does this paper attempt to address?