Machine Learning Based Secured Reliable Routing Protocol to Predict and Prevent Byzantine Attack in MANET

The emergence of Mobile Adhoc Networks (MANETs), in which multiple nodes join alongside one another to provide the necessary real-time services, has garnered increased attention from academics due to recent advancements in wireless communication. The wireless connectivity and decentralized architecture of MANETs, however, make constructing secure routing a challenging task. As a result, a machine learning-based Secured Reliable Routing Protocol (ML-SRRP) is suggested in this study to predict and prevent Byzantine attacks in MANET. In order to prevent unwanted packets with routing information being sent to fake destinations (in order to prevent routed loops), the nodes that serve as intermediates in a network were initially selected using an authority evaluation method at every single node. The node-specific metrics Number of Hop Count (NHC), Residual Energy Level (REL) and Link Validation were used for this. In order to effectively identify attacks and isolate hacked nodes within the network, an intrusion detection system is necessary. In this research, the Ad-hoc On Demand Distance Vector (AODV) routing MANET protocol is used to introduce the Byzantine attacks. Support Vector Machine (SVM) is used to train and test the system to identify illegally nodes in the network. The simulation results show that the proposed ML-based SRRP outperforms current methods in terms of dependability and throughput. Moreover, there is a reduction in delay, routing overhead, and packet loss ratio. R was used for testing and training with the trace file that was taken from the NS-2 simulation, which was run to carry out Byzantine attacks.