Journal Homepage: www.ijcst.org

Abstract – The complexity of attacks on computer systems is 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 paper presents the investigations carried out on different neural network structures using a number of algorithms for intrusion detection. Also this paper 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 paper 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 Data mining (KDD) dataset to experiment the performance of ESN in classifying the Local Area Network (LAN) intrusion packets