Pre-processing Algorithm for Rule Set Optimization Throughout Packet Classification in Network Systems

With recent advancement in various networking technology, many field packet classifications have evolved from traditional classification so as to classify large rule sets. Most of the previous algorithms provide excellent performance when rule set was small. As rule sets grew in size, performance degraded due to lack of memory and do not have enough processing capabilities to route incoming packet at such a high rate. Packet pre-processing is one of the most important aspects of classification, as it will increase the throughput as well as improve the search performance. The proposed method mainly focuses on pre-processing of pre-defined rule set used during classification. In proposed approach, double hashing technique is to optimized memory usage for high throughput. Proposed algorithm implemented on Xilinx ISE design suite 14.2 with 10000-50000 rules. Simulation results shows that the memory consumption is only three fourth compared to existing approaches.