This research aims to optimize the traffic signal cycle and the green light time per traffic signal cycle at ramps and intersections in arterials to maximize the passing traffic volume and minimize the delaying traffic volume in freeway corridors. For this purpose, we developed the MATDYMO (multi-agent for traffic simulation with vehicle dynamics model) and validated it with comparison to commercial software, TRANSYT-7F, for an interrupted flow model and to URFSIM (urban freeway traffic simulation model) for an uninterrupted flow model. These comparisons showed that MATDYMO is able to estimate the traffic situation with only incoming traffic volume. Using MATDYMO, ramp metering and traffic signal control can be optimized simultaneously. We extracted 80 sampling points from the DOE (Design of Experiment) and derived each response from MATDYMO. Then, a neural network was adopted to approximate the objective function, and simulated annealing was used as an optimization method. There are three cases of the objective function: maximization of the freeway traffic volume, minimization of the delay of ramps and arterials, and the satisfaction of both cases. The optimization results showed that traffic flow in freeway corridors can be maintained to a steady stream by ramp metering and signal control.