Sustainable cyber-physical VANETs with AI-driven anomaly detection and energy-efficient multi-criteria routing using machine learning algorithms

Cyber-physical systems have improved modern transportation by allowing vehicles and road systems to communicate through Vehicular Ad Hoc Networks (VANETs). Existing anomaly detection approaches often struggle with high false-positive rates, poor adaptability, and significant computational demands, compromising their real-time efficacy and scalability. To address these problems, this research presents an Anomaly Detection using Machine Learning Algorithms (AD-MLA) framework that employs a Random Forest model to accurately detect abnormal activities. The framework encompasses feature selection, data clustering, and an energy-efficient routing strategy that incorporates node energy, signal strength, hop count, and link stability. Evaluations demonstrate that AD-MLA reduces false alarms, improves detection accuracy, and operates with lower energy and computational requirements. It offers a smart, rapid, and efficient security system for real-time VANET environments, rendering it appropriate for transportation systems characterised by high reliability and safety. By integrating a Random-Forest-based anomaly detector with intelligent feature selection and an energy-efficient routing method that accounts for residual energy, signal strength, and link stability, the suggested framework systematically addresses these challenges. This approach delivers 95.33% accuracy, 96.09% recall, 94.25% computational efficiency, and 91.45% resource-use efficiency. This effectively addresses the scalability, latency, and energy challenges that previous systems have faced in incorporating blockchain technology and deep learning architectures.