A Novel Approach for Pattern String Matching in Intrusion Detection System

Abstract

String matching algorithms are critical for web usage devices which screen the packets and moves on the basis of load. Intrusion detection or prevention systems, web cleansing, anti-virus and junk preventive systems all have a necessity for effective algorithms when managing string matching scenarios. Some of the algorithms viz. Aho-Corasick, WuManber, Rabin-Karp, Commentz-Walter, Bit parallel etc. entails lot of space and time. AhoCorasick algorithm is built on the basis of finite state machines. On the shadow of KnuttMorris-Pratt, Commentz Walter developed an algorithm which in turn uses finite state machines. Bit parallel algorithms like shift-or takes the help of varied computer words ie. the registers in CPU, to compare the efforts. Rabin Karp implements hashing for discovering matching pattern strings in a given text. Wu-Manber looks text in chunks than of every character coalescing the work of Aho-Corasick and Boyer-Moore. Our algorithm reduces character comparisons and consumption of memory on the basis of graph evolution structure and search procedure by means of dynamic linked list. Theoretic study and trial results demonstrates that the projected work is much enhanced in terms of space and time consumption, when compared with previously known pattern-matching algorithms.