Abstract: Aimming at the traditional intrusion detection system is difficult to adapt to the increasing amount of data demand for real-time processing capability, this paper uses of sliding window and the data stream clustering technology to design a clustering algorithm based on sliding window data streams, and build the IDS network security defense model based on the algorithm. The validation of model simulation proves that the network security defense model is able to adapt to the high-speed network intrusion detection requirements.

Key words: data streams, mining algorithms, intrusion detection, network security

摘要： 针对传统入侵检测系统难于适应日益增长数据量对实时处理能力的需求问题，运用滑动窗口、数据流聚类技术，设计了基于滑动窗口数据流聚类算法，并构建了基于该算法的IDS网络安全防御模型。通过对该模型仿真验证，证明该网络安全防御模型能较好地适应高速网络的入侵检测需求。

关键词: 数据流, 挖掘算法, 入侵检测, 网络安全