S. Lin, B. De Schutter, Y. Xi, and H. Hellendoorn, "Model predictive control for urban traffic networks via MILP," Proceedings of the 2010 American Control Conference, Baltimore, Maryland, pp. 2272-2277, June-July 2010.
Model Predictive Control (MPC) is an advanced control strategy that can easily coordinate urban traffic networks. But, due to the nonlinearity of the traffic model, the optimization problem of the MPC controller will become intractable in practice when the scale of the controlled traffic network grows larger. To solve this problem, the nonlinear traffic model is reformulated into a model with only linear equations and inequalities. Mixed-Integer Linear Programming (MILP) algorithms can efficiently solve the reformulated optimization problem, and guarantee the global optimum at the same time. Moreover, the MILP optimization problem is further relaxed by model reduction and adding upper bound constraints.