A Clustering Algorithm Based On Density Gravity And Application In Intrusion Detection

Intrusion detection, by definition, is the process about discovering intrusion behaviors. Using a number of key points from the computer network system to collect and analyze information, and found the behavior with violating security policy and signs of accusation, then automatic response. It not only detects intrusion behavior from outside, but also monitors the user's unauthorized activity. It is considered to be the second gate after firewall, and is one of the core technologies of network security. In recent years, intrusion detection has become a field of comprehensive subject. Data mining, nerve network, machine study techniques have been applied to the research of intrusion detection.With the development of computer and network technologies, intrusion technology has been more and more diverse and complex; it takes more pressure to the intrusion detection system. Data mining technology provides an effective means to solve this problem, using data mining method as a data analysis technology of intrusion detection. It may extract as many as possible hidden security information from vast amounts of security events to discover the intrusion. Therefore, combination of data mining and intrusion detection technology, it can increase the processing of the mass data to intrusion detection system. Cluster analysis is one of the many usually data mining method. It is a typical unsupervised study technology, and can establish directly intrusion detection model on the data set with not mark or abnormal data. It has great application value to improve the detection power of intrusion detection system.In this paper, it main researches an improved clustering algorithm based on density gravity. This algorithm can automatically determine the number of clusters in the target data set, and can find arbitrary shape clusters, filter the "noise" data, improve the clustering results. Then, devising the experimental model, and test the results of the clustering algorithm of density gravity used to the intrusion detection.Finally, this paper gave an experiment about clustering algorithm of density gravity used in intrusion detection, using KDDCUP 99 data set. The result of experiment proved that the algorithm had a high detecting rate and a low distorting rate.