Parallel Optimization And Research Of String Matching Algorithm In Intrusion Detection System

With the rapid development of network technology, the network security problems have become increasing serious and attracted people’s attention, detection of network content has become an important component of the network security system. As the core technology of network security checks, string matching algorithm is facing enormous challenges in handling massive amounts of data and a variety of application needs. Due to improving performance of the classical serial string matching algorithm in intrusion detection system has become very small, the parallel instead of serial is the only effective way to solve this problem in complex network environment. With the risen and development of multi-core technology, the parallel algorithm based on multi-core has become a research focus.This article is based on multi-core processor platform and string matching algorithm research, carried out realization、optimization and improvement of parallel of string matching algorithm.(1) The experimental analysis of existing single-mode exact string matching algorithm. In this paper, experiment for the pattern string with different length respectively in the size of 32 and 64 character sets, obtain the algorithm with highest performance in general case.(2) Through the analysis of the existing string matching algorithm, this paper proposes an improved Horspool algorithm based on the Horspool algorithm which has good performance in practice. The improved algorithm increases the average moving distance of the window and improves the matching efficiency.(3) This paper uses data decomposition method to design and implement the parallelism version of the improved Horspool algorithm and Shift-Or algorithm based on dual-core and quad-core processor platforms, and implements this parallel algorithm with the multi-threaded development tool OpenMP.(4) The performance of the parallelized code is analyzed with the current popular high performance analysis tool VTune. Then this paper optimize the degree of parallelism, load balancers, cache hit rate and other issues of the parallel algorithms, then apply them to specific string matching experiments, analyze the performance of the matching speed of the two parallel algorithm on a dual-core and quad-core processor platforms, speedup and other performance with experimental data.The parallel method on multi-core processor platform which is proposed in this paper can be used in other string matching algorithm to improve their performance.