Packet classification based on the decision tree with information entropy

Packet classification is indispensable for the next-generation routers targeting at the complete integration of advanced networking capabilities, which include differentiated services, memory access control, policy routing, and traffic billing. The classification method based on decision tree is advantageous in its structure and high efficiency, so it is suitable for real-time packet classification. A heuristic method is proposed based on the information entropy to build the decision tree more balanced considering the time complexity and the space complexity. It is suitable to solve rule subset uneven phenomenon and meets the requirement of big data with diverse data formats. The simulation results show that the algorithm can classify the packets quickly compared with previously described algorithms and has relatively small storage requirements.