Watchdog malicious node detection and isolation using deep learning for secured communication in MANET

Mobile Ad-hoc Networks (MANETs) are wireless networks formed dynamically by connecting or leaving nodes to and from the network without any fixed infrastructure. These categories of wireless networks are susceptible to different attacks based on their dynamic topological structure. Due to this, security is a primary constraint in MANETs to preserve communication between mobile nodes. A Deep Neural Learned Projective Pursuit Regression-based Watchdog Malicious Node Detection and Isolation (DNLPPR-WMNDI) technique is proposed and modelled in this paper to improve the security feature of MANETs. The newly proposed DNLPPR-WMNDI technique initially selects the neighbouring nodes by applying the projection pursuit regression function. In multicasting, the route paths are established through the intermediate node with the help of control commands named RREQ and RREP. After then, Watchdog Malicious Node Detection and Isolation (WMNDI) technique is applied to detect malicious nodes based on the data packet forwarding time. Basically, a malicious node is affected by a node isolation attack. For better communication, a malicious node is isolated from the network and multicast routing is carried out by selecting the next neighbouring node and this improves the communication security. Simulation is done for the developed technique based on different performance metrics.