Pattern matching is commonly required in many application areas and bioinformatics is a major area of interest that requires both exact and approximate pattern matching. Much work has been done in this area, yet there is still a significant space for improvement in efficiency, flexibility, and throughput. This paper presents a hardware software co-design of Aho-Corasick algorithm in Nios II soft-processor and a study on its scalability for a pattern matching application. A software only approach is used to compare the throughput and the scalability of the hardware software co-design approach. According to the results we obtained, we conclude that the hardware software co-design implementation shows a maximum of 10 times speed up for pattern size of 1200 peptides compared to the software only implementation. The results also show that the hardware software co-design approach scales well for increasing data size compared to the software only approach.
@article{arxiv.1403.1317,
title = {Hardware software co-design of the Aho-Corasick algorithm: Scalable for protein identification?},
author = {S. M. Vidanagamachchi and S. D. Dewasurendra and R. G. Ragel},
journal= {arXiv preprint arXiv:1403.1317},
year = {2014}
}