Misbehaviour Detection of 5G-Connected Vehicles Using Deep Learning

Connected and Autonomous Vehicles (CAVs) have the potential to revolutionize transportation and road safety. For CAVs to communicate amongst themselves, they use Vehicle-to-Vehicle and Vehicle-to-Infrastructure (V2V and V2I) communication to share information among vehicles and infrastructure, leading to better hazard detection and safer driving. However, the increased connectivity and use of various sensor systems also pose security risks, making it crucial to develop misbehaviour detection (MBD) mechanisms to enhance the security and safety of CAVs. The following paper proposes a novel deep-learning approach for MBD in a CAV system running on a 5G network. The proposed approach aims to improve the accuracy and efficiency of misbehaviour detection, addressing the growing security concerns in CAVs. A two-model approach will be outlined, designed to learn and understand the expected behaviour of real cars, making the system robust to unknown attacks and future threats. The system will also show an ability to understand the context in which a car is operating, to improve MBD accuracy. Recommendations will be outlined regarding improved security measures through BSM verification by neighbouring vehicle BSMs and using 5G network information.