Towards scalable optimal traffic control

This paper deals with scalable control of traffic lights in urban traffic networks. Optimization is done in real time, so as to take into account variable traffic demands. At each cycle of the traffic lights, the optimization concerns time instants where each traffic light starts and ends its green phase: this allows to describe both the duty-cycle and the phase shifts. First, we formulate a global optimization problem, which can be cast as a mixed-integer linear program. To overcome the complexity of this centralized approach, we also propose a decentralized suboptimal algorithm, whose simplicity allows online implementation. Simulations show the effectiveness of the proposed strategies.