A high throughput and memory-efficient regular expression matching on FPGA

Regular expression matching engine is a key component in many content inspection based network applications. But it is facing two issues: low speed and state explosion. In order to address these issues, we propose a novel architecture for high throughput and memory-efficient Regular Expression Matching Engine (REME) on FPGA in this paper, which is based on Instruction Finite Automaton (IFA). REME can reduce over 95% states and take O(n) time. We demonstrate Snort's 1083 regular expressions in single Altera Stratix IV EP4SGX530NF45C2ES, it can achieve 11.3 Gbps.