Estimating Daily Start Times of Periodic Traffic Light Plans from Traffic Trajectories

In recent years, the wealth of available vehicle location data from connected vehicles, cell phones, and navigation systems has been introduced. This data can be used to improve the existing transportation network in various ways. Among the most promising approaches is traffic light optimization. Traffic light optimization has the potential to reduce traffic congestion, air pollution and GHG emissions. The first step in such optimization is the understanding of the existing traffic light plans. Such plans are periodic but, in practice, often start every day at arbitrary times, making it hard to align traffic trajectories from various days toward the analysis of the plan. We provide an estimation model for estimating the daily start time of periodic plans of traffic lights. The study is inspired by real-world data provided, for instance, by navigation applications. We analyze the accuracy of such computations as a function of the characteristics of the sampled traffic and the length of the evaluated time period.