Learning-Based Security Technique for Selective Forwarding Attack in Clustered WSN

Abstract

Selective forwarding attacks in WSN can damage many mission-critical applications, like military surveillance and forest fire censoring. In such attacks, malicious nodes most of the time functions like regular nodes, but sometimes drop sensitive packets selectively, like a packet recording the dissimilar power' activity, making it more difficult to identify their malicious intent. The current selective forwarding attack detection schemes, randomly select checkpoint nodes, available in-between nodes within a forwarding route, which are responsible for producing acknowledgments for each received packet. In this paper, the complete sets of nodes are differentiated into three different types based on their functionality as Inspector Node (IN), Cluster Head (CH), and Member Nodes (MN). The newly considered node as IN is considered to overhear all of the activities of the Cluster head, as CH is the most compromising node in the complete cluster, and in the case, if the CH is attacked then the complete cluster stops working in the network. The IN is trained based on certain rules and predefined parameters which analyses if the CH or MN is malicious or not and considers the required action. NS2 is considered for the simulation of the proposed methodology and also for the validation of the proposed work. In the proposed methodology, two different stages are considered as detection and correction, which works to tackle the attacks and also considering the system efficiency almost. As in the proposed methodology, the effect of the attack is minimized which increases the QOS and also better data transmission.