文档介绍：摘要
入侵检测系统(Intrusion Detection System,IDS)就是通过分析计算机系统的网络连接数据和审计日志记录等相关数据信息,来检测入侵行为的系统。入侵行为是对目标系统的非授权访问或以降低目标系统性能为目的,以获取非法利益的行为。目前已有的入侵检测系统存在检测方法复杂,检测代价大,对新型入侵检测率低的缺点。粗糙集理论通过对信息决策系统数据中不可分辨关系的处理,能有效精简数据属性特征维数,提取更简单的规则以降低数据分析的代价。因此,粗糙集理论很适合网络入侵检测系统处理数据量大,实时性要求高的特点,适用于入侵检测系统的研究。
本文首先介绍了入侵检测系统的基本概念和研究现状,指出了传统的入侵检测方法存在的问题。接着介绍了粗糙集相关理论知识,在数据预处理环节提出了一种新的连续属性离散化方法,在保证信息不丢失的前提下,更方便粗糙集的处理。在粗糙集基本理论的基础上,不仅运用命中集的概念,提出了一种新的属性重要度的构成方法,以降低计算复杂性,还提出一种基于基因抽取的遗传算法,用于粗糙集属性精简的运算,提高了收敛速度。随后介绍了现有的入侵检测系统的模型,并在此基础上,结合粗糙集理论,设计了一种基于粗糙集属性精简的入侵检测系统模型。最后通过仿真实验,验证了本文提出的方法的可行性和有效性。
关键词:入侵检测粗糙集命中集属性重要度遗传算法
Abstract
The Intrusion Detection System(IDS) can detect intrusions by analysing puter system'work connection data and audit diary record. At present, the existing IDSs plex methods, big operating prices and low detecting rate to new styles of intrusions. The rough set theory can make use of the indiscernible relation of data in information decision systems to reduce the attribute features. It can reduce price of data analysis and simplify the rule extraction by reducing the attribute dimensions. Because of the large quantity and large number of the features work connection data have, the rough set theory is very suitable to the research work intrusion detection systems to reach the real-time target.
This thesis firstly introduced the IDS's basic concept and the research situation at present, and pointed out the defects of traditional intrusion detecting methods. Then we introduced the rough set theory and its related knowledge. Based on the discernibility matrix we proposed a new representation of attribute importance with the hitting set concept applied. And then, we proposed a ic algorithm based on the gene extraction to reduce the attribute features. Then we introduced the existing IDS's model, based on rough set theory we designed a new IDS model. The simulation experiment confirmed the feasibility and validity of the method this thesis proposed.
Keyword:Intrusion Detection,Rough Set The