Many-Objective Optimization for the Traffic Signal Synchronization Problem

Traffic Signal Optimization addresses this challenge by optimizing traffic signal cycles using optimization algorithms and traffic simulators, such as SUMO. These simulators allow the modeling of traffic scenarios and the evaluation of multiple traffic quality metrics, which naturally formulate the problem as a Many-Objective Optimization Problem (MaOP). Although previous studies have explored multi-objective formulations of traffic signal control, few investigate the use of many quality metrics and broader analyses of MaOP algorithms. This paper models the traffic signal synchronization problem as a MaOP and presents a performance analysis of MaOP algorithms. Two multi-objective evolutionary algorithms, NSGA-II and NSGA-III, and an objective reduction technique are applied using the jMetal framework integrated with SUMO. An experimental study on two traffic scenarios analyzing  correlations between quality metrics and the behavior of optimization algorithms.  The results show that NSGA-III outperforms NSGA-II in most scenarios and objective reduction is effective only in the simpler scenario.