Data-driven Soil Moisture Sensing with mmWave Radar

Yujie Zhuang, Yang Zhao, Jie Liu

Abstract: Existing soil moisture sensing methods require either plugging probes into soil or burying sensor nodes or aluminum plate in soil. These methods either have mobility and maintenance limitations or have risks of battery leakage. To address these issues, we propose to build a soil moisture sensing dataset using millimeter-wave radar for non-invasive and non-contact moisture sensing of shallow soil. Our millimeter-wave radar soil moisture dataset includes 20 different moisture levels, filling the gap in mmWave radar-based soil moisture sensing datasets. Furthermore, we propose a mm-SoilNet model, which uses the data-driven approach to extract soil moisture features from radar data to estimate the volumetric water content of soil. Our initial experimental results show that the system achieves a mean absolute error of 5.506%.