Packet classification with filter categorization

In next generation networks, routers use the technique of packet classification to classify incoming packets into multiple forwarding classes. The classification is based on predefined filters which indicate the requirements of quality of service or security handling. To achieve scalable packet classification, numerous algorithms based on hash tables have been proposed. These algorithms focus on reducing the number of accessed hash tables to improve the search performance; however, their performance improvement is mainly affected by filter characteristics. We analyze the search procedures of hash-based algorithms and propose a scheme of adaptive filter categorization to improve the search performance. Our scheme stores the filters in different data structures according to their characteristics. Therefore, the search performance could be improved by eliminating the worst cases in previous algorithms. Our scheme is evaluated with filter databases of varying sizes and characteristics, and the experimental results demonstrate that our scheme is feasible and scalable.