Novel Intrusion Detection System for MANETs using Hybrid Improved Grey Wolf Optimizer based deep recurrent neural network model

Abstract The advent growth of wireless networks and their applications in different fields has been more prominent in the previous few years and specifically mobile ad-hoc networks (MANETs) has gained the importance from consumers and researchers due to their reliability and sustainability. MANET is widely employed for communicating and sending and receiving packets in a network without any requirement of specific hardware structure, due to which these are employed in different sectors. Because of its wide applicability, there exist numerous challenges in handling MANETs, especially with respect to network security. Intrusion is one of the key security problem faced in MANETs during the transmission of data packets within the communication system. The occurrence of malicious node in MANETs shall lead to removal of information packets during data transfer and thus intrusion detection system (IDS) are successfully designed to handle the node behaviours and identify the malicious nodes in the network and their behaviours. Henceforth, this research study develops a novel IDS model using the proposed improved grey wolf optimizer (ImGWO) hybridized with that of the deep recurrent neural network (DRNN) model for predicting and observing the behaviour of malicious nodes in MANETs. The developed novel hybrid ImGWO-DRNN model is applied on the KDD Cup 1999 datasets and training and testing of the proposed IDS technique is studied. Evaluation metrics was set to analyse and validate the proposed IDS technique and the results prove the superiority of this ImGWO-DRNN intrusion detection system over the previous techniques from literature for the MANETs.