Decision-Making under Deception at Intersections: A Bayesian Game Approach

Connected Autonomous Vehicles (CAVs) offer substantial potential to improve urban mobility by enabling safer, more efficient, and environmentally sustainable transportation. However, deceptive behaviours-such as transmitting false information or acting unpredictably-pose significant risks to safety and system reliability, particularly at intersections where rapid decisions must be made under uncertainty. Unlike existing work that treats deception solely as a security threat, this paper models deception as a strategic choice using a Bayesian game-theoretic framework. Our approach introduces independent additive penalties for efficiency, comfort, and safety that adjust based on the probability of deception and relative vehicle positions, thus balancing strategic interactions between leading and lagging vehicles. The framework is evaluated via simulations under fixed, adaptive, and Nash equilibrium-based yielding scenarios. Results demonstrate that equilibrium-driven strategies significantly mitigate the adverse effects of deception while enhancing fairness and overall utility in connected autonomous traffic systems.