作者: J. Jiang , X. Wang , K. He , B. Liu
关键词: Throughput (business) 、 Finite-state machine 、 Parallel processing (DSP implementation) 、 Deterministic finite automaton 、 Parallel computing 、 Pattern matching 、 Deep packet inspection 、 Computer science 、 High memory 、 Intrusion detection system 、 Speedup 、 Regular expression
摘要: Multi-pattern matching is a key technique for implementing network security applications such as Network Intrusion Detection/Protection Systems (NIDS/NIPSes) where every packet inspected against predefined attack signatures written in regular expressions (regexes). To this end, Deterministic Finite Automaton (DFA) widely used multi-regex matching, but existing DFAbased researches have claimed high throughput at an expenses of extremely memory cost. In paper, we propose parallel architecture DFA called Parallel (PDFA), using multiple flow aggregations to increase the with nearly no extra The basic idea selectively store modules which can be accessed and explore potential parallelism. cost our system both average cases worst analyzed, optimized evaluated by numerical results. evaluation shows that obtain speedup about 0.5k 0.7k k number under synthetic trace compressed real statistical case, compared traditional DFA-based approaches.