Quality of Forecasts for Signal Timings at Traffic Lights

Abstract

The report documents the development of a concept for the evaluation of forecast quality for signal timings at traffic lights. Nowadays, modern traffic light systems are capable of using C-ITS technologies to share information with other road users not only about their current signal state but also about predicted future signal states. To meet the requirements for future-oriented and high-performance C-ITS services, such as the GLOSA service (Green Light Optimal Speed Advisory), a concept for evaluating the forecast quality of such prediction data has been developed. It comprises three central evaluation steps: a signal control dynamics index that enables the comparability of individual evaluations across different junctions and signal control systems; a separate check of the integrity and plausibility of the forecast data; and a substantive evaluation of forecast quality based on the analysis of signal timing data. A final overall rating in the form of quality levels can be derived from these individual evaluations. The proposed concept provides project partners and stakeholders from cities and local authorities with an initial basis for a systematic assessment of the quality of forecast data and prediction methods. It is intended to be tested in greater detail and further developed in practical C-ITS applications in the future, thereby establishing a reliable tool for evaluating the forecasts of traffic signal timings.