基于硬件的快速正则表达式匹配方法

High-Speed Matching Approach of Regular Expression Based on Hardware Support

网络应用中基于正则表达式特殊模式识别技术是比较新颖的一门学科。基于NFA的方法速率较慢,而基于DFA的执行算法会耗费大量的存储空间。文中提出一种分组模式处理方法,将序列庞大的正则表达式编译为少量的DFA,然后利用多核处理器来并行地处理分组模式,在不明显增加内存耗费的情形之下增加了正则表达式的匹配速率。

Regular expression-based special pattern identification in networking applications is a new subjuct. The traditional NFA-based approach is slow in speed while the DFA-based implementation would expend large quantity of memory. To address this, a grouping algorithm is proposed, which compiles a large set of regular expressions into a small number of DFA, and then the multi-kernel processor, is used to parallelly process the grouped patterns, thus dramatically improving She regular expression matching speed without significantly increasing the memory expense.