The project focused on developing an intelligent, networked sensor system for urban traffic monitoring and vehicle counting, with a particular emphasis on ultrasonic sensing technology. A custom ultrasonic platform was designed to transmit, receive, and pre‑process acoustic signals, enabling real‑time object detection. Initial algorithms employed the DBSCAN clustering method to group reflected signal samples that exceeded a threshold amplitude and were spatially close. To achieve optimal detection, a large‑scale parameter search was conducted on a cluster of 20–50 computers, with MATLAB used to compare the resulting clusters against manually labelled video data. This iterative optimisation process highlighted the sensitivity of DBSCAN to external interference and the computational burden of exhaustive parameter tuning.

To reduce complexity and improve robustness, the sensor data were re‑interpreted as pulse‑Doppler radar signals. Moving vehicles produce a Doppler frequency shift, which was detected using a bank of matched filters. The envelope of the matched filter output was analysed; the difference between successive sample blocks was calculated, and the root‑mean‑square error (RMSE) of these differences was used to identify peaks corresponding to vehicle presence. The direction of motion was inferred from the sign of the frequency shift. Sensor fusion, combining the outputs of two ultrasonic units, further enhanced detection accuracy. The final algorithm was ported to an embedded digital signal processor and tested on a street in Berlin, demonstrating reliable vehicle counting under real‑world conditions.

Parallel to algorithm development, hardware integration efforts produced compact printed circuit boards that could be embedded directly into streetlight fixtures or mounted on pole structures. These boards incorporated LED driver modules and were certified under ENEC and CE standards. The project also established a modular, gateway‑based communication architecture that supports V2X (vehicle‑to‑everything) connectivity. End‑to‑end network prototypes were built using custom ICE Gateway devices, and an application programming interface was defined to allow external partners to deploy device‑specific software through a model similar to an app store. User‑friendly dashboards and portals were developed for device management and data visualization, enabling operators to monitor traffic flow and sensor health in real time.

Collaboration among the partners was coordinated over the 2016‑2018 project period, with the final report covering work up to the end of 2018. The funding source was the I2EASE programme (grant number 16EM00147). ICE Gateway GmbH led the hardware and gateway development, integrating the ultrasonic platform into streetlight prototypes and managing the V2X communication stack. RWTH Aachen University contributed algorithmic expertise, particularly in applying digital image‑processing techniques to ultrasonic data and conducting the DBSCAN optimisation studies. Osram and several LED manufacturers (Selux, RetroFit, Intec, Helux) supplied lighting fixtures and facilitated the integration of the gateway electronics into existing streetlight infrastructure. Together, the partners defined clear work‑package boundaries, aligned on use‑case requirements, and delivered a suite of prototypes that demonstrate the feasibility of networked ultrasonic traffic monitoring within urban lighting systems.