A Real-Time Implementation of an Edge-Assisted Collision Warning System for Urban Intersections

Communication-based collision warning systems (CWSs) play a pivotal role in enhancing safety, particularly during adverse weather conditions or when obstacles obstruct drivers’ and sensors’ visibility. In this study, we implement and validate a practical and scalable CWS tailored for urban inter- sections, incorporating machine learning-based trajectory and collision predictions, Cooperative Awareness Message dissemination, and Decentralized Environmental Notification Message alerts. The validation process entails testing the system on a cellular testbed, exploiting Open Air Interface, and leveraging realistic traffic data to closely replicate dense urban scenarios. Through extensive experimental testing, our findings demonstrate the system’s efficacy in proactively identifying dangerous situations well in advance, even within densely populated urban environments, thereby enabling timely evasive maneuvers.