Genetic vs. particle swarm optimization techniques for traffic light signals timing

A good controlling for the traffic lights on the network road may solve the traffic congestion in the cities. This paper deals with the optimization of traffic light signals timing. We used four different heuristic optimization techniques, three types of Genetic algorithm and particle of swarm algorithm. Techniques were applied on a case study of network road which contains 13 traffic lights. We used SUMO (Simulation of Urban MObility) to simulate the network. Heuristic optimization techniques themselves need to be calibrated. Calibrating them using the real problem is time consuming because simulation is computation demanding. We tried to calibrate them using a function that is assumed to have similar response surface but lighter computation demand, then use the calibrated technique to optimize the traffic light signals timing. After some comparing processes of optimization results, we discovered that one type of GA and PS at determined parameters are more suitable to produce the minimum total travel time.