A Case Study of Evaluating Traffic Signal Control Systems Using Computational Experiments

A new traffic signal control system (TSCS) evaluation method that uses computational experiments based on artificial transportation systems (ATSs) is proposed in this paper. Some basic ideas of the method are discussed, i.e., generating reasonable travel demand, modeling the influence of environment, and designing communication interface. Using a 30-day computational experiment on ATSs, a case study is carried out to evaluate three TSCSs, which are implemented using fixed-time (FT), queue-based responsive (QBR), and adaptive dynamic program (ADP) algorithms, respectively. Aside from normal weather, three types of adverse weather, i.e., rain, wind, and fog, are modeled in the computational experiment. After analyzing aggregate data and detailed operating record, reliable evaluation results are obtained from this case study. Furthermore, several interesting phenomena are observed in this case study, which have yet to be noticed by previous work.