RCO based Multi Layer IDS in MANETs with Collective Fluctuation Measure

Abstract

Intrusion Detection system (IDS) has a vital role in identifying the misbehavior in network. By its malicious behaviour the performance of the network gets deteriorated. The two major types of wireless networks that are infrastructure less in its operation is Mobile Ad-hoc Networks and WSNs. There is a huge challenge in the research of MANETs as there are limited resources available, due to which there is need for an Intrusion Detection System. There is multiphase scheme adopted in the proposed method, the first one is to utilize the security location leader mode based algorithm which can be incorporated based on the location of the node. This leader will fetch the exact data from the other network nodes within its radio range.

Further, the decision tree classifier is used by the Security leader node. The leader node collects the Abnormality Index from every node in the network at different positions.

The second method is called multi-layer IDS which includes two layers of malicious node detection. The first-layer uses security location leader based approach that has dedicated sniffers that collects the data from its neighboring nodes and calculates a new parameter called ‘Rightly Classified Occurrence’ which is then forwarded to the leader node at constant reporting intervals. This is considered based on taking the variance of Rightly Classified Occurrence in smaller size for malicious nodes and larger size for normal nodes. Based on this, a new measure called Collective Fluctuation (CF) is calculated in the second layer. The result show better when there is low speed or velocity and has high transmission capacity when the scenario compared to low transmission and high velocity.