A Fast Algorithm for Packet Classification with Reasonable Memory Usage

Packet classification is one of the basic techniques for many applications such as routers, firewalls, and intrusion detection systems and so on. With the development of the Internet, however, packet classification has already become a performance bottleneck in network infrastructures. A great many algorithms for packet classification have been proposed. Although most of these algorithms run fast, they consume very large amount of memory. This paper presents a fast algorithm for packet classification with reasonable memory usage called RCC (rule component cuttings). Similar to HiCuts (Hierarchical Intelligent Cuttings) which is one of the best-known algorithms for packet classification, RCC is also based on a decision tree to categorize an incoming packet. In spite of this, there exists one important difference between them. That is, HiCuts divides a dimension into equal-sized intervals yet RCC cuts it into unequal-sized intervals, which gives rise to decreasing the amount of memory. Experimental results show that the time of HiCuts is smaller than RCC´s but the amount of memory HiCuts consumes is much larger.