초록

용어보기논문에서 용어와 풀이말을 자동 추출한 결과로, 시범 서비스 중입니다. AI-Helper

Abstract ▼ AI-Helper

As one of the most challenging tasks in designing Internet routers, packet classification is required to achieve wire-speed processing for every incoming packet. Algorithms based on area decomposition are efficient packet classification algorithms since they can examine 5 different fields (source IP...

As one of the most challenging tasks in designing Internet routers, packet classification is required to achieve wire-speed processing for every incoming packet. Algorithms based on area decomposition are efficient packet classification algorithms since they can examine 5 different fields (source IP prefix, destination IP prefix, source port number, destination port number, and protocol type) at the same time. As well-known area-based algorithms, an area-based quad-trie(AQT), a hierarchical intelligent cutting(HiCuts), and an HyperCuts algorithm have been studied. The AQT algorithm requires the smallest memory amount among the area-based packet classification algorithms, but it does not provide high-speed search performance in case when many rules belong to the crossing filter set of a node. The HiCuts algorithm provides high-speed search since the number of rules compared with each input can be controlled by a limited number. However, the HiCuts is not practical because of huge memory requirement caused by rule replication which cannot be avoided in constructing the HiCuts decision tree. This paper proposed a new algorithm which primarily constructs an area-based quad trie(AQT) and then applies the HiCuts tree for crossing filter sets with more number of rules than a pre-defined number. Hence the proposed algorithm shows reasonable performance in both the search speed and the memory requirement since the number of rule comparison is reduced compared to AQT algorithm as well as the rule replication is limited by a small set of rules belonged to the constructed HiCuts trees.

주제어