用于网络入侵检测的免疫学习子系统  

An Immune Learning Sub-system Used in the Network Intrusion Detection System

作　　者：张四海[1] 罗文坚[1] 曹先彬[1] 王煦法[1] 

机构地区：[1]中国科学技术大学计算机科学技术系,安徽合肥230026

出　　处：《小型微型计算机系统》2003年第8期1441-1444,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金 ( 699710 2 2 )资助 ;安徽省自然科学基金 ( 0 0 0 43 10 6)资助 ;博士点基金 ( 19990 3 5 82 6)资助

摘　　要：网络入侵检测当前面临的主要问题是如何迅速有效地检测出未知模式的入侵 .借鉴生物免疫系统中的自进化学习机制 ,我们设计了一种免疫算法 ,该算法以生物免疫的自我非我识别为基础 ,并进一步引入免疫学习机制以提高算法对入侵模式识别的效率和正确率 .算法由四个紧密联系的模块组成 :基因库进化模块、亲和度变异模块、非我选择模块和免疫记忆模块 ,四者形成一个有机的整体 .本文介绍了免疫算法的具体细节 ,并完成了相应的验证实验 .The main problem of current network intrusion detection is how to recognize the intrusion of unknown patterns rapidly and efficiently. Inspired by the self evolution learning mechanism in the biological immune system, we designed an immune algorithm, which is based mainly on the SELF/Non _SELF recognition. We also used the learning mechanism to improve its efficiency and preciseness for reconnizing intrusion patterns. The algorithm consisted of four modules: gene recombination module, affinity mutation module, negative selection module and gene evolution module, and these four components formed an integral whole. This paper introduced the details of the immune algorithm and accomplished corresponding experiments. The experiment results given showed that this algorithm had a good ability to recognize unknown patterns.

关 键 词：网络入侵检测 人工免疫 基因库进化 亲和度变异 非我选择 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]