DESIGN- OF AN EFFICIENT MOBILE AGENT BASED INTRUSION DETECTION SYSTEM

Abstract

An intrusion is an attempt to compromise a computing resource by exploiting the vulnerabilities present in it. Intrusion Detection System is a piece of software that detects any such attempt and raises an alarm. A distributed system is a network of interconnected autonomous computers. Intrusion Detection in a Distributed System is highly challenging because of the quantity of the data and the dynamic nature of the network. Mobile Agents have been used to design Intrusion Detection Systems for Distributed Environment and have been proven to have several advantages over conventional systems, like, overcoming network latency, reducing network load, autonomous execution, scalability and adaptability. An Intrusion Detection System analyzes data which is specified by a number of features to detect intrusions. Reduction in the number of features can lead to faster detection with even greater accuracy because of the presence of irrelevant and redundant features. The existing systems mainly use hierarchical systems with single or multi-step detection with only one detection technique at each step, and also they have not considered the benefits of reducing the size of the dataset with feature reduction. In this research work, we propose a mobile-agent based Intrusion Detection System which uses feature selection to reduce the size of the dataset for faster detection, two different types of classification mobile agents, and a meta classification agent to combine the results of the two to give the best output. We then compare the empirical results obtained with the results of the other existing techniques to assert the accuracy of our proposed mechanism