A Study on the Application of Data Stream Clustering Mining through a Sliding and Damped Window to Intrusion Detection

With an ever-greater increase in network bandwidth, network speed and network traffic, network attack techniques are constantly changing and improving, making it formidable for the traditional network security defense measures to keep pace with this challenge. In this paper, a theoretical analysis is made first of both the traditional intrusion detection and the data stream mining, and then, a research is conducted into a network security defense technique based on the integration of data stream mining and intrusion detection, thereby coming up with an algorithm in the light of data stream clustering mining through a sliding and damped window. And this algorithm is applied to the intrusion detection systems so as to approach the traditional problem of inadequate real-time intrusion detection. Through analysis and simulation, it turns out that the algorithm has a lower requirement for operating environment but a higher clustering quality, thus facilitating good reference to improvement in the performance of intrusion detection.