SVM BASED APPROACH FOR INTRUSION DETECTION IN MANET

Abstract

A mobile ad-hoc network (MANET) is a collection of mobile devices having routing capabilities to reach to the destination. Co operation of all the nodes in the network is essential for efficient performance of the network. But cooperation of all the nodes cannot be assumed in such a dynamic network in which nodes are joining and leaving very frequently and topology changes
randomly. Some of the nodes can be compromised that adversely affects the performance of the network. Different types of
attacks can be performed at various layers of the protocol suite An Intrusion detection system is required to accurately detect
attacks and isolates compromised nodes in the network. In this work black hole attack has been introduced using Ad-hoc On
Demand Distance Vector (AODV) as Routing MANET protocol and the system is trained and tested using Support Vector Machine (SVM) to detect the malicious nodes in the network. The simulation was done in NS-2 to carry out black hole attack and trace file obtained is used as dataset for training and testing purpose using R.