多特征融合的入侵检测

Intrusion Detection Based on the Fusion of Multiple Features

从人认识事物的特点出发,对认识过程进行抽象,把它用到入侵检测系统中,提出了多特征融合的入侵检测方法。在此方法的实现中,以特征元素为基本单元,形成特征规则,采用基于可信度的不精确推理,判断系统被入侵的可能性。还讨论了模式的形成,给出了实现模型。此方法能有效提高计算机系统抵御入侵及自身免疫的能力。

In this paper, a pattern recognition approach to network intrusion detection based on the fusion of features is proposed, which matches with the process by which human distinguish objects. Feature rules which is consist of feature elements and uncertainty reasoning based on measure of belief are introduced to make the decision of ID. The pattern forming and the model of such an approach are outlined.