Comparison of signalized junction control strategies that use localization probe data

This paper presents a simulation test bed and a methodology for evaluating urban signalized junction control algorithms that use localization probe data from all vehicles in the local area. The simulator is based on SIAS Paramics microsimulation software with bespoke software modules built on top for automatic network generation, localization data processing and signal control. Also presented are results from tests carried out using the simulation test bed to evaluate localization strategies. The tested strategies use a hierarchical structure of auctioning agents. Results from tests on an isolated signalized junction indicate that the performance of the auctioning agent algorithms compare favourably with the MOVA algorithm using inductive loop data. Results are also presented for tests on a twin junction where strategies are synchronized. These show a significant improvement in performance through synchronization.