高效匹配引擎的设计与实现

The design and implementation of high-performance matching engine

针对软件模式匹配引擎速度慢、占用系统资源大的问题,提出了一种基于Bloom Filter的FIBF结构,结合FPGA NIOSII微处理器(MUP),设计了软硬核相结合的匹配引擎方案。引擎通过FIBF过滤过滤掉大部分正常数据,将过滤出的可疑字符串输入到NIOSII软核进行精确匹配,两者之间通过一个指针产生器连接,整个系统以流水线方式工作。FIBF结构资源消耗低,n-FIBF并行处理,系统引擎可以达到较高的吞吐率。实验表明,在使用相同资源的情况下,本方案系统吞吐率要优于其他算法。

According to the problem that the pattern matching engine based on software having slowly matching-speed and occupying large system resources, this paper proposed a FIBF structure based on Bloom Filter. Using the FPGA NIOSII microprocessor（ MUP）, a matching engine combination of hard and soft core is designed. Though FIBF, the engine filtered out most of the normal data, then sent the suspicious string to the precise matching module. The FIBF module and the precise matching module were connected together by an index generator. The whole system was performed in pipeline method. The FIBF has low consumption of hardware resource. N-FIBF process in parallel can achieve high system throughput. Simulation results show that this engine is superior to other algorithms by using the same resources.