Application And Research On Intrusion Detection System Based On Improved K-means Algorithm

With the continuous development of network technology and application field expands unceasingly, Internet has become the main part of daily life. However, its opening, makes it facing the enormous risks of the security. Intrusion detection is the firewall, encryption, identity authentication and access control measures to protect the safety of tradition such as a new security technology. It can not only real-time detection from outside intrusion activities, also supervise internal user unauthorized activity, and the computer and network resources malicious use behavior identification and response. Effectively the traditional network safe protection technology. But with the complication of computer system and huge quantity of network data the system of the intrusion detection demands a higher level.。Based on the above, according to the background of intrusion detection technology, from the false positives and detection rate of two important indicators, using wave data mining methods will improve the K-means algorithm to the intrusion detection system, in order to improve the performance of the intrusion detection system, and the simulation experiments. The paper studies are as follows:From the intrusion detection development present situation, and introduces the intrusion detection, the intrusion detection system, puts forward the concept and principle of the intrusion detection system is the problem, and the future development direction. According to the theory of clustering analysis, this paper introduces the traditional algorithm, means-K K value from the clustering center, the choice and noise outlier with three aspects of traditional K-MENS algorithm was improved, and achieved good effect.Finally, the algorithm to the improvement of the intrusion detection system, and puts forward the intrusion detection model, and the simulation results show that the processing of data. Experiments prove that the improved means to improve the data K-the clustering effect, based on the improved algorithm of K--means intrusion detection system reduces the error detection rate and the error rate and improve the quality of intrusion detection.