Publication Details
Optimalizace architektury NFA-Split
NFA, FPGA, regular expression matching
Number of signatures used in intrusion detection systems is steadily growing in connection with the development of computer networks. This entails the necessity of reduction of utilized FPGA resources for implementation of non-deterministic finite automata used to accelerate regular expressions matching. The article discusses possibilities of optimizing the NFA-Split architecture. Two optimization are described in detail. The first optimization is a replacement of determinization algorithm with an algorithm with quadratic time complexity. The achieved acceleration is 4-39 fold. The second optimization is the introduction of inner alphabets in deterministic part of the NFA-Split architecture. Reached reduction of the utilized BRAM memories is up to 98%.
@INPROCEEDINGS{FITPUB10387, author = "Vlastimil Ko\v{s}a\v{r}", title = "Optimalizace architektury NFA-Split", pages = "81--86", booktitle = "Po\v{c}\'{i}ta\v{c}ov\'{e} architektury a diagnostika PAD 2013", year = 2013, location = "Plze\v{n}, CZ", publisher = "University of West Bohemia in Pilsen", ISBN = "978-80-261-0270-0", language = "czech", url = "https://www.fit.vut.cz/research/publication/10387" }