Increasing the Efficiency of TCAM-based Packet Classifiers using dynamic Cut Technique in Geometric Space

Packet classification is one of the main processes that often run on network processors. In hardware implementation of packet classification algorithms, Ternary Content Addressable Memories (TCAMs) are used to implement parallel search and process packets rapidly. In classifier architecture, first, decision tree is created and classifier rules are distributed among its leaves. In second stage, rules are included in different blocks of TCAM corresponding to leaf of the tree structure. In this study a new dynamic algorithm is offered to select the best bits for cutting in representation of rules in geometric space to distribute them equally and reduce their duplication in the decision tree. Efficiency of the proposed architecture which uses dynamic cuts has been compared with recent architectures. Comparing results shows that the proposed method can distribute rules in TCAM block more balanced than recent architectures. Therefore, memory and power consumption requirements are reduced considerably.