基于决策树的启发式流分类算法目标是建立结点数目尽可能少、树深度尽可能小的数据结构,从而获得较优的时空性能,据此理论提出的基于参数评估的可调节式流分类算法(PEA),一方面沿袭目前主流的决策树类流分类算法思想;一方面引入性能参数的概念,并采取调节参数权值的方式获得性能最佳的数据结构.测试结果表明,相同条件下本算法对比同类算法能获得更优的性能结果.
Heuristic packet classification algorithm based on decision tree is aimed to classify packets with minimal time and space requirements. An adjustable algorithm based on parameter evaluation is presented. It follows the idea of popular packet classification algorithms, introduces the conception of performance parameters, and adiusts weights of these parameters to aquire data structure with the best performance. Simulations show that, compared with other algorithms of the same kind, a good improvement can be obtained when using the new algorithm.