Enhancing security in vehicular Ad hoc networks: A novel approach using DSFLA, SACVAEGAN, and OAEF

Abstract

Important types of MANETs include Vehicular Ad hoc Networks (VANETs), where wirelessly linked automobiles form a network. The safety of VANETs is a pressing concern that is currently being addressed, as a large number of lives can be lost due to even a single security failure. Intrusion Detection Systems (IDS) are active in VANETs to detect any intrusion and ensure security. Traditional IDSs struggle to keep up with the proliferation of sophisticated, pattern-based assaults. The IDS scans the network to identify any malicious nodes. In this study, we present a novel optimum feature-selection method called the differentiated shuffled frog-leaping algorithm (DSFLA). To further enhance classification performance, the paper introduces a Self-Attention-Based Conditional Variational Autoencoder Generative Adversarial Network (SACVAEGAN) to produce virtual samples and enrich the training data. An enhanced artificial electric field (AEF) technique, called opposition-based AEF (OAEF), is used to find appropriate hyper-parameters for the tuning process. We conclude by testing the suggested approach on the Car Hacking dataset. This research not only examines the efficacy of the suggested framework but also evaluates the efficiency of common categorization tasks. The suggested technique has been shown to be an efficient IDS, with experimental findings on the Car Hacking dataset demonstrating its outperformance of the current state-of-the-art deep learning representations.