Distributed model predictive control of freeway traffic networks: A serial partially cooperative approach

In this paper, a new distributed model predictive control (MPC) scheme for freeway traffic control is proposed. It is aimed at reducing the communication efforts and the computation times in a large network. This new algorithm can coordinate a large number of on-ramps throughout a freeway network in a partially cooperative scheme. The communication is performed between neighboring on-ramps in a special serial fashion and with three different proposed cooperative schemes. The computation time is much less than that of existing distributed MPC approaches in the literature, while achieving a performance close to the one of the centralized MPC method. To evaluate the performance of the proposed partially cooperative schemes, a freeway network case study is selected and the problem of coordination between several on-ramps is solved using different methods from a centralized approach to a fully decentralized one. The obtained results show a significant decrease in the total computation time with respect to the centralized and fully cooperative schemes, while maintaining a close distance to the optimal objective function obtained from the centralized case. Furthermore, the performance of the proposed partially cooperative MPC method is evaluated in the case of incidents in the network.