OBSERVE: Blockchain-Based Zero Trust Security Protocol for Connected and Autonomous Vehicles (CAVs) Data Using Simple Machine Learning
Bo Sullivan, Junaid Ahmed Khan · 2024
Connected and Autonomous Vehicle (CAVs) can proactively share future trajectories with each other and coordinate for safe navigation and efficient route planning. However, for a vehicle to rely on a nearby vehicle's trajectory data for realtime navigation decisions, it needs to trust data shared by neighboring vehicles. There exist no such distributed method to do so, and therefore, we propose OBSERVE, a lightweight trust model for nearby vehicles to endorse each others' trajectories in realtime using blockchain. Vehicles self-organize to endorse each others trajectories through consensus, where, each vehicle verify nearby vehicles trajectory by predicting the corresponding trajectory, compare results with peers, and reach consensus on its truthfulness. We employ simple machine learning for vehicles to predict neighbors trajectories instead of the computationally heavy vision based prediction algorithms, thus leveraging times-tamped coordinates towards an energy efficient endorsement. OBSERVE is validated on realistic data and has shown to achieve higher prediction accuracy with lower computation overhead.