Design and analysis of ANN-based echo state network intrusion detection in computer networks

Abstract

The complexity of attacks on computer systems are increasing rapidly. The current network is complicated due to the high throughput and the multi-uniformity of actions. Intrusion detection is a process of monitoring the various computer networks and systems for violations of security and this can be automatically done with the help of an intrusion detection system. An Intrusion Detection System (IDS) is a critical component for secure information management. IDS plays a major role in detecting and disrupting various attacks before cooperating with the software. This study presents the investigations carried out on different neural network structures using a number of algorithms for intrusion detection. Also, this study proposed an Echo State Network (ESN) structures for intrusion detection. The proposed algorithm has faster convergence and better performance in IDS. The objective of this study is to implement the ESN algorithm and compare with other neural network algorithms in a networked environment. The performances of different methods have been implemented and compared using the Knowledge Discovery and Datamining (KDD) dataset to experiment the performance of ESN in classifying the Local Area Network (LAN) intrusion packets. © Medwell Journals, 2012.