High performance pattern matching using Bloom-Bloomier Filter

In this paper, we propose a high performance architecture based on the combination of Bloom Filter and Bloomier Filter (BBF) to enhance the speed of pattern matching process on Clam Antivirus (ClamAV) database. BBF maintains small on-chip memory, low number of fault positives and can indicate which patterns are the candidate matches. The implementation results on low-cost Altera Cyclone II show that our architecture can handle 43,491-characters of ClamAV pattern set with only 9.5 bits per character and achieve a throughput of 1 gigabit per second (Gbps). As compared with previous systems, our memory utilization is far better up to 73%.