Misbehavior Detection in Internet of Vehicles: A Multi-Dimensional Survey

<title>Abstract</title> <p>Cooperative Intelligent Transportation Systems (C-ITS) are central to next-generation Internet of Vehicles (IoV) and IoT-enabled mobility, facilitating real-time data exchange among vehicles, roadside units (RSUs), and cloud-edge infrastructure. While these systems improve safety and traffic efficiency, their open, distributed design exposes them to insider threats, data falsification, and emerging attacks driven by Generative AI (GenAI). This survey proposes a multi-dimensional taxonomy for C-ITS misbehavior detection, covering detection layers, methodologies, deployment contexts, and adversary models. It maps techniques such as federated learning, blockchain, cross-layer semantic reasoning, and GenAI-resilient frameworks to real-world vehicular applications. Our analysis shows that privacy preservation, accuracy, latency, and scalability are vital for Artificial Intelligence (AI)-enabled attacks and plausibility checks. This survey concluded with future challenges in C-ITS, including privacy preservation, and emphasized lightweight, explainable, and interoperable solutions. The outcomes of this survey inform the design of a GenAI and a future attack-resilient defence system against C-ITS misbehaviour attacks.</p>