Research of dynamic self-adaption intrusion detection model based of clustering
Yan Hong-yin · Jisuanji gongcheng yu sheji · 2013
Traditional intrusion detection based of Clustering research mostly through improving algorithm to improve the effect of intrusion detection,it is usually not efficient in time and memory.Algorithm parameters are decided by artificial attempts,it is difficult to ensure the parameters is the best and dynamic.The problems are solved by a new intrusion detection model,the preprocessing method adapt for the algorithm is used in the algorithm.The environment which K-means applied in is made good use of,useful information about intrusion is used in the execution of the algorithm.so it has a quick convergence speed,and the problems K-means itself has are solved.A dynamic intrusion detection model is established through setting up an reasonable dynamic method to confirm The begining center vectors and Radius threshold parameters.The intrusion detection model is useful verified by experiment,and it can be used to detect a special intrusion.