ABSTRACT

Security issues in MANET are of primary importance due to its higher vulnerability to attacks compared to wired networks. Though attack prevention mechanisms using authentication and encryption become the first line of defense, newer types of attacks cannot be detected by such systems. A second line of defense is necessary to detect and respond to intrusion. In MANETs the attack can occur within the network from one of the nodes or could be perpetuated from outside the network through the gateway. Attacks from external node not only affect the target node but also the participating nodes in the network. In this paper we evaluate Intrusion Detection Classification Techniques on a MANET gateway connecting to a wired network. Bayesian and Decision tree induction techniques are evaluated and results presented.