数据平面和控制平面相分离的网络入侵检测系统

Research on Efficient Network Intrusion Detection Based on Data Mining Algorithm

本文提出一种具有数据平面和控制平面分离的网络入侵检测系统,实现了高可扩展性。结合考虑单分类支持向量机和软间隔支持向量机的优点,提出了一种基于增强支持向量机的入侵检测方法,以此准确高效地区分恶意入侵数据流与正常数据流。在仿真实验部分,使用恶意软件产生恶意数据集,并利用该数据集来验证系统的有效性。

This paper presents a network intrusion detection system with data plane and control plane separated,which can achieve high scalability.By combining the advantages of one-class SVM and soft-margin SVM,an enhanced support vector machine based intrusion detection is proposed to classify malicious intruded and normal data flows accurately and efficiently.In the simulation experiment,we use malicious software to generate malicious data set,which is used to verify the effectiveness of the proposed system.