Research On Packet Classification Algorithm And Its Application In Firewall

With the development of computer network and communication technology, the high speed and strong time effectiveness characteristics of network make the equipments such as security gateways and routers present the network with higher demand. The speed of packet classification algorithms which is the core algorithm of these equipments, is becoming the bottleneck of internet transmission. A excellent packet classification algorithm is also the basement of internet different services. So do some research on the high-effect packet classification technology has deep influence on the development of internet in the future.Packet Classification is the technology which based on the domain of packet head and in accordance with the rules in the database to classify the packet. In general, the problem of packet classification is also the problem of point location in the hyperspace. In this dissertation we introduce packet classification algorithm research background, the definition of packet classification algorithm, the research status and its application in the firewall meanwhile we also analyse some kinds of packet classification technology which based on many kinds of different principles and do some deep research on the benefit and disadvantage of the AFBV algorithm. In this article, we propose an enhanced algorithm EAFBV in which we use the theory of tuple space to regroup the rule database and make the rules with same condition in one single tuple to accelerate the positioning of the rules in the initialization time, meanwhile we use using single folding replace multi folding to reduce the number of the folding time and effectively improved the efficiency of the algorithm.In this dissertation we build a simulation platform to test time efficiency between EAFBV and AFBV under the Linux operation system. The experiment results show that EAFBV algorithm is more efficiency in the initialization time and in the average search time than AFBV algorithm. So it is suitable for using in the scalable rule database.