Consider a road network with severe traffic congestion. A set of efficient strategies can be addressed to simultaneously minimize road users’ travel times and maximize possible increase in travel demands. Due to the non-differentiability of the perturbed solutions in equilibrium constraints, non-smooth optimization models for delay-minimizing and capacity-maximizing signal setting problems are established. In this paper, we propose a new bundle subgradient projection approach to solve the signal setting problems with global convergence. Various control policies are numerically tested one against another. Numerical results disclose that the proposed approach has successfully solved the delay-minimizing and capacity maximizing signal setting problems and achieved significant performance in cost reduction when compared to other alternatives.