Model Predictive Control is applied to control and coordinate large-scale urban traffic networks. However, due to the large scale or the nonlinear, non-convex nature of the on-line optimization problems solved, the MPC controllers become real-time infeasible in practice, even though the problem is solvable in theory. In this thesis, we mainly focus on the solutions for increasing the real-time feasibility of the MPC controllers for large-scale urban traffic networks.