An Empirical Review of Scenario-Aware Traffic Signal Control Models from A Statistical Perspective

Abstract

Traffic signal deployments have increased many folds due to an exponential increase in road-traffic in both urban and rural areas. Thus, modern-day town planning agencies focus primarily on using optimized-traffic-signal-control models that can assist in reducing congestions and maintaining smooth traffic flows. Researchers have discussed design of various machine-learning based traffic signal control models, and each of these models vary in terms of their operational & functional characteristics. Moreover, these models have highly variant efficiency levels, due to which it is difficult to select optimal models for scenario-aware use cases. Due to these variations, it is difficult for city planning agencies to identify optimal models for their contextual use cases. To reduce this difficulty in model selection, a survey of some of the recently proposed models for traffic signal control is discussed in this text. This discussion evaluates the reviewed models in terms of their application-specific nuances, functional advantages, deployment-specific limitations, and contextual future scopes. Based on this evaluation, it was observed that bioinspired models, pre-emptive analysis models, and non-linear processing models outperform others. After referring this discussion, researchers will be able to identify optimal traffic-signal-control models for their functional use cases. To further simplify the process of model selection, this text also compares the reviewed models in terms of their computational complexity, deployment cost, fairness of control, scalability, and congestion control efficiency levels. After referring to this comparison, researchers will be able to contextually identify traffic-signal-control models suited for their deployment-specific use cases. This text also combines these parameters in order to identify models that can be deployed with low cost, have low complexity, are capable of improving fairness of control, with high scalability and congestion control efficiency performance under high traffic scenarios.