Optimized Multi-Model Intrusion Detection Scheme for Internet of Vehicles

Intelligent vehicles are increasingly connected to the external world, enabling functionalities like vehicle-to-vehicle (V2V) and vehicle-to-everything (V2X) communication and data transmission. However, improving connectivity also increases the vulnerability to cyber-security threats, raising concerns about data privacy, unauthorized access, and potential system interference. Due to insufficient authentication and encryption, Controller Area Network (CAN) and external network gateway in Internet of Vehicles (IoV) is highly susceptible to network attacks, necessitating robust intrusion detection systems (IDSs). In this paper, an optimized multi-model intrusion detection scheme is proposed for IoV using multi-model stacking ensemble framework, combining k-nearest neighbors (KNN), bagging classifier, and random forest (RF) models. The framework optimizes classification hyper parameters to enhance intrusion detection. Simulation and experiment results show that the proposed multi-model framework can effectively detect the intrusion by 99.9% on average and maintains a lower time complexity, thereby enhancing overall intra-vehicle and external network efficiency.