Queue Length Estimation from Connected Vehicles with Low and Unknown Penetration Level

Queue length estimation has been a long-standing problem in transportation systems as it provides an important component for the design, operation, and performance monitoring of signalized intersections. In this paper, we present a novel estimation algorithm based on trace data from connected vehicles. In contrast to existing algorithms, our algorithm only requires a very low level of penetration rate (~1 %) for connected vehicles. As such, it is already applicable given the current level of penetration in practice. Moreover, it is agnostic to the actual value of the penetration rate or any other information about an intersection. We provide verification of our algorithm via numerical simulations. We demonstrate the application of our algorithm using real-world data for four intersections.