An Efficient Multi Level Intrusion Detection System for Mobile Ad-Hoc Network Using Clustering Technique

Abstract

Mobile Ad-Hoc Network (MANET) is constructed using autonomous self-disciplinary nodes that communicate and exchange information through wireless medium. VM based IDS identify the attacks in the Virtual Machine Level with isolated properties of data center at the cloud. This is efficient only at the data center level i.e. infrastructure level. These problems are addressed by the proposed multi – level IDS for MANET using clustering technique. It identifies the black-hole attack from the external level to internal level. The pattern of the internal as well as external attacks are extracted and stored into the knowledge base for further analysis. The nodes are clustered and select an arbitrary node as a cluster head. The topology also monitored for maintaining consistency over the detection. The MANET packets are compared with the knowledge base to detect the malicious packets. The malicious node can also be eliminated from the network. Various modern IDS tools are analyzed with large set of attacks in multiple levels in order to maintain high reliability. Different algorithms are compared with proposed IDS in performance evaluation metrics such as IDS rate, Positive Rate and alarm rate and so on. The proposed IDS provides high accuracy when compared to existing algorithms in all levels