A secure data routing protocol based on encryption and hybrid deep learning in MANET

Mobile Ad hoc Networks (MANET) permit mobile users to communicate with each other when fixed infrastructure is not possible. However, the vulnerability risks are the major problem in MANET routing. Hence, ResneXt Quantum Dilated Convolution Neural Networks (RQDCNN) is designed for secure data communication in MANET. The routing, secure node identification and bi-filtering phases are considered. In the routing, the route discovery and route reply are performed. To avoid route list modification, node's address is encrypted by Rivest-Shamir-Adleman (RSA) algorithm. The secure nodes are determined using the RQDCNN in the secure node identification, which is the integration of ResneXt and Quantum Dilated Convolutional Neural Networks (QDCNN). Finally, the important nodes are filtered in the bi-filtering phase using network-based parameters. The RQDCNN obtained better throughput of 9.746 Mbps, delay of 0.635 ms and Packet Delivery Ratio (PDR) of 98.55% with black hole attack.