A Packet Classification Algorithm based on Improved Decision Tree

Since the packet classification algorithm has its great importance and extensive applications, lots of scholars have done tons of researches on the algorithm. This paper presents a new method of a packet classification algorithm based on improved decision tree. In the proposed algorithm, multi-point segmentation and redundancy deletion are adopted to compress the depth of decision tree. A new data structure and the interval binary search are introduced to optimize the time performance. Simulation results show that, the average decision tree depth of MPS is approximately 60% of HyperSplit, and the memory access of MPS is 10% less than HyperSplit.