A study of distributed control strategies for urban traffic

This study investigates a set of distributed systems for urban traffic control. The distributed systems considered herein include TUC, demand-based, and max-pressure con- trollers. The strategies are tested on a medium sized... [ view full abstract ]

This study investigates a set of distributed systems for urban traffic control. The distributed systems considered herein include TUC, demand-based, and max-pressure con- trollers. The strategies are tested on a medium sized network in Bloomsbury area in Central London. The distributed systems are also compared with the optimal control solution derived from centralised optimisation. The results show that the centralised control policy can generally outperform the distributed ones in terms of delay reduction, which is expected. However, the former takes much longer time to compute than the distributed strategies. Moreover, the study shows that the benefit of centralised control is indeed limited when the level of traffic is low or when it is close to the physical capacity of the network. This study provides insight on choosing different control architectures under different circumstances.