基于状态约束的大规模正则表达式匹配算法

摘要

By analysis of state explosion in deterministic finite automata DFA, a novel algorithm Group2-DFA based on state constrains was proposed to reduce the memory usage. With the state constrains, states in NFA were classified into several groups. Group2-DFA introduces two-level classification and merges NFA and DFA together to a hybrid FA construction. The experiments show that Group2-DFA can reduce memory usage efficiently and keep high throughput with a small increase of memory reading time. With 300 regex rules, Group2-DFA can cut 75%states and achieve 1Gbps throughput.%通过观察不确定有限自动机NFA到确定性有限自动机DFA的转化过程，分析内存增长的原因，提出了一种基于状态间约束关系的正则表达式匹配算法Group2-DFA。Group2-DFA通过两级分组，利用状态间的约束关系，将原始NFA转化为NFA和DFA的混合结构。实验表明，在保持一定处理速率的前提下，Group2-DFA能够有效地减少内存占用。在300条规则下，Group2-DFA吞吐率能够达到1Gbps，并且减少约75%的状态数。