Design of a Multi-agent Based Intelligent Intrusion Detection System

Abstract

Considering some employees in the department abuse their privilege for personal gain through the local network and the host computer, in this paper, we present a multi-agent based intrusion detection system named MAIIDS (multi-agent based intelligent intrusion detection system). The learning agent module in this model can self-adjustingly learn network based audit data and host based audit data with more than one technique of data mining, such as the artificial neural network, the association rules and so on. The learning agent can produce rules, and the detection agent can detect audit data according to these rules and response to them. A lot of experiments indicate that this system has very high self-adapting ability, intelligence, and expansibility. Now MAIIDS has been applied to several security departments and has received good reputation