Application of the proximal center decomposition method to distributed model predictive control