Research on Rapid Detection Algorithm Based on Property Choosing Tree
Neng Shan Feng, Zhong Ming Yang · Advanced materials research · 2012
The construction method used by detection engine Snort-NG based on ID3 decision tree has the problem of excessive memory occupancy. The idea that the test properties are chosen according to the gradation of rule property in network protocol stack was presented in this paper; that is, the property of link layer first determined, and then network layer and transport layer. The atomicity of the value of these properties were preserved and the values of these properties were treated as a whole. The results of experiment showed that the occupancy of memory was much less in the state of non-trivial property being very common with this approach.