DL-Enabled Road Traffic Monitoring Services Using 6G mmWave Channel State Information

Future wireless networks will integrate sensing functionality to improve communication performance and offer services for verticals. An example use case of 6G sensing could be in the context of future intelligent transportation systems (ITSs) aiming for versatile benefits, including environmental and socioeconomic factors. To this end, accurate road traffic information acquired in real-time by ubiquitous cellular networks will be paramount to impose instant measures and develop data-driven long-term solutions. This paper showcases the wireless radio channel as a privacy-preserving environmental sensor intrinsically affected by its surroundings, facilitating edge intelligence for road traffic monitoring. Notably, we consider millimeter-wave (mmWave) deployments owing to their propagation and communication system characteristics for sensing, i.e., low penetration and diffraction combined with directional beamforming antennas and high bandwidth. We employ real-world mm Wave channel traces gathered in an urban road traffic scenario to demonstrate its suitability for two use cases well suited for future 6G sensing services. Utilizing machine learning (ML) and deep learning (DL) models, we demonstrate road user classification accuracy of up to 99.8 % and 97.8 % for binary and three-class traffic counting, respectively. In the second use case, we outline the feasibility of mm Wave-enabled estimation of passing vehicles' moving direction with up to 92.3 %.