FCVN: Future Communications in Vehicular Networks With Hybrid Machine Learning Model for Detecting Vehicular Attack

Abstract

Intelligent transportation systems (ITS) rely heavily on Future Communication in Vehicular Networks (FCVNs), which allows real-time communication between vehicles and infrastructure to enhance traffic efficiency and road safety. However, the integrity and dependability of ITS can be compromised by several security risks. This study uses the Vehicular Reference Misbehavior (VeReMi) dataset, a benchmark dataset with various vehicle attack scenarios, to offer a Hybrid Machine Learning (ML) framework for detecting vehicular attacks on ITS. Using performance parameters like accuracy, precision, sensitivity, $$-score, specificity, and FPR, the hybrid ML models including K-Nearest Neighbors (KNN) and Naive Bayes (NB) have been assessed and compared with state-of-art approaches. With a detection accuracy of 97.85% much greater than the accuracies documented in comparable studies, the results show that the proposed hybrid ML model performs better than existing techniques. The results highlight how crucial it is to use a hybrid model to improve vehicle security and guarantee the secure and effective functioning of FCVNs in practical situations.