A real-time network-level traffic signal control methodology with partial connected vehicle information

•Development of two traffic state estimation algorithms for partially connected transportation networks.•Developing a real-time methodology for traffic signal control with partial connected vehicle information.•Achieving significant improvement in traffic operations compared to existing approaches.•Even at 10% CV market share, the number of completed trips increased by 3.2–3.5%. This paper presents two algorithms to estimate traffic state in urban street networks with a mixed traffic stream of connected and unconnected vehicles and incorporates them in a real-time and distributed traffic signal control methodology. The first algorithm integrates connected vehicles (CV) and loop detector data to estimate the trajectory of unconnected vehicles based on car-following concepts. The second algorithm converts the temporal point vehicle detections to a spatial vehicle distribution on a link. The signal control methodology utilizes either algorithm to estimate traffic state on all network links at a time, optimizes the signal timing parameters over a prediction period constituting several time steps, implements the optimal decisions in the next time step, and continues this process until the end of the study period. We applied the methodology to a real-world case study network simulated in Vissim. The results show that both algorithms are effective under a wide range of CV market penetration rates in all tested demand patterns: at 0% market penetration rate, the proposed methodology reduced travel time by 2% to 10% and average delay by 7% to 20% compared to the existing signal timing parameters and traffic demand. At a 40% penetration rate, the proposed algorithms reduced travel time by 27% to 33%and average delay by 50% to 61% compared to the existing signal and demand pattern in the case study network. Similar trends were found for all other demand patterns tested in this study.