Improved K-medoids Algorithm For Intrusion Detection

The rapid development of information technology(it) is beyond our imagination, and has been widely applied to every field of life, our daily life is inseparable from the computer now, even the appearance of many new technology products, they are very much like the workings of a computer, such as smart phones and tablets. We use these high-tech products communicate, learn, and the existence of the network, it even went to shopping, eating, travel, etc. A list of details of life. The subsequent is a severe test to the information security problem. And the types of network virus growth speed is very fast. In their daily lives more and more people are stolen, a growing number of hackers using illegal methods, access to property, Caused great damage to people's normal life, but due to the limitation of ability and each person's work and study direction is not the same, we can't require everyone to use the Internet will protect your computer security, only through the computer system itself, improve on the network security performance, to prevent the happening of the intrusion.People have to seek good method, to defense the invasion, but the hacker technology are also rising, so the traditional firewall static passive defense way faced with big challenges. And intrusion detection of active defense, just the good makes up the defect of the traditional technology.In this paper, based on the intrusion detection technology, with data mining clustering analysis as the breakthrough point, seeking to deal with large data environment data mining intrusion detection technologies. Clustering analysis is emerging with the development of computer technology research of a research direction of data mining technology, it will be a large amount of data, classified according to certain rules, make similar data gathered into a class, in order to distinguish.And the characteristics of clustering to accord with the requirement of intrusion detection to the algorithm, the invading behavior is relative to the first daily network behavior is less, and can according to different intrusion characteristics, group similar points out, clustering integration. So this paper mainly studies the improved k center clustering algorithm combined with intrusion detection methods, the article includes the following several aspects:First, on the intrusion detection system and system in this paper, the clustering analysis method, analyzes the existing clustering method is applied to intrusion detection in large data environment when the advantages and disadvantages.Second, introduce the relevant technology of the network intrusion detection, clustering analysis and k center clustering algorithm in the data store environment problems intrusion detection, and based on the original k- medoids algorithm was improved, will improved k- medoids algorithm applied to intrusion detection.Third, the use of a recognized the KDD CUP99 dataset in the current intrusion detection experiment, finally the experimental results show that the improved algorithm can better adapt to the big data environment intrusion detection, and greatly improve the efficiency of intrusion detection, also can effectively reduce the time space complexity.