A Multiattribute Measurement Algorithm for Packet Classification
Many network application technology need the algorithm for multi-dimensional packet classification, for example ,network security ,load balancing ,router policy, QoS etc. Considering the levels of multiattribute packet classified are excessive and traverse rule table times without number for matching classification rule, so efficiency is lower. A packet classification algorithm based on decision tree is put forward in the paper. As compared with some traditional packet classification matching algorithms, because three data are adopted including information gain, information gain ratio and Gini to solve attribute selection measurement, accuracy and matching efficiency are both advanced obviously.
M. L. Pen and A. M. Huang, "A Multiattribute Measurement Algorithm for Packet Classification", Applied Mechanics and Materials, Vols. 52-54, pp. 168-173, 2011