Distributed Acoustic Sensing of Precipitation with Deep Learning and Scalable Engineering on Edge Hardware

We introduce a distributed acoustic sensor network designed for real-time rainfall detection. This network offers high spatial and temporal resolution and utilizes MEMS-based vibration transducers mounted on photovoltaic modules, which serve as sensing interfaces. Embedded microcontrollers enable scalable IoT connectivity, low‑power operation, and minimal maintenance, supporting integration into photovoltaic infrastructure. Synchronized data acquisition and flexible data management enhance edge inference with resource efficient deep neural networks. The development process is based on a framework that supports robust data flow from data acquisition to on-device analysis and application-tailored databases. This platform aims to improve evidence‑based decision-making in urban planning, agriculture, and infrastructure management, while maintaining low operational costs and supporting large-scale, adaptable deployment. By providing high‑resolution precipitation situational awareness, this sensing approach can contribute to civil protection and emergency preparedness, particularly for extreme rainfall and flood risk mitigation.