Decentralized Intersection Control Using Bayesian Game Theory

Future smart cities are expected to have efficient control of vehicular traffic to provide satisfactory mobility and transportation of people and goods. Reliable and efficient control schedule design for signalized intersections is needed to alleviate vehicular congestions and improve the overall road network management. In the present paper, we propose an approach based on game theory to design a decentralized intersection traffic controller, able to adaptively react to changing traffic conditions and minimize the waiting time of the cars in queue. We adopt a Bayesian dynamic game framework, which is able to improve the existing state of the art alternatives by reducing the amount of data exchanged from monitoring roadside units. Moreover, we also introduce a tunable sharing factor that is a design element available to the traffic planner controlling the priority of access and allowing for prioritization. Finally, the proposed solution is extensively evaluated via simulation in different scenarios.