Parallel GPU-based hybrid string matching algorithm

Abstract

String or pattern matching is an essential part of most computer applications that are widely used in text processors, Internet-based search engines and computer security. A key concept of string matching is identifying the existence of a pattern with m characters in a much longer text string. String matching operations are basic to many algorithms such as indexing algorithms, search algorithms and bioinformatics algorithms that are used in molecular biology, adopted pattern matching concept in their programming structure. Although the mathematical and algorithmic calculations of pattern matching algorithms are simple, they are still the main reason of a large ratio of the computational load in many applications This research proposes to parallelize a hybrid string matching algorithm called Maximum-Shift algorithm, by using CUDA device (GPU) to increase its speed up during the matching process. The parallel Maximum-Shift algorithm shows slight improvement over the sequential version in terms of running time, speed up rate and percentage of performance gain.