Research And Implementation Of Fast Packet Classification Algorithm

The development of Internet application necessitates next–generation router's abilityto support those functions such as QoS(Quality of Service), Network Intrusion DetectionSystems (NIDS), Measurement of traffic, Accounting and Billing, Loading Balance andControl Congestion etc. Although implementation of these functions vary mostly, they allneed packet classification.So packet classification has extensive application in networktechnology field. It provides foundation for many important network technologies andis closely associated with many aspects of network, such as network control, networksecurity, network management and network performance. Especially, it isindispensable to next-generation router to support many new applications. The speedand function of packet classification directly influence on the performance of thosenetwork technologies.At first, by analyzing the existing algorithm and characteristic of classifier, the authordiscusses principle and approach about how to design packet classification algorithm.Meanwhile,the author also discusses the future of packet classification and it's application.Then the author discusses emphatically a new packet classification algorithmnamed CPI(compress, partition and index ). This packet classification algorithms is basedon PCBNP, RFC and regional partition algorithm and by designing two storage datastructure, compress partition rule table and index list table, this algorithm can support tenthousand multi-dimension rules, and have fast rate of packet classification, low expectedstorage complexity, can be used in IPv6 and be implemented in software or hardware.Compared with existed classification algorithms, this algorithm is better than those existedclassification algorithms under ten thousand about multi-dimension rules, so it can be usedin wide range.