Dynamic Soil Moisture Estimation Using BLE RSSI Signals: A Machine Learning-Based Framework for Real-Time Monitoring and Flood Detection

Accurate soil moisture monitoring is vital for precision agriculture, flood prediction, geoscience and environmental management, but existing In-Ground Sensor Systems (IGSS) are limited by high cost, complexity, power consumption, and scalability issues. This paper presents a novel, low-cost, non-intrusive hybrid solution that estimates real-time soil moisture and detects floods using Bluetooth Low Energy (BLE) RSSI signals. The approach combines in-situ sensing for deep soil layers with remote sensing for shallow layers, overcoming key limitations of traditional IGSS. In this approach, the BLE communication infrastructure serves a dual purpose: transmitting deep soil sensor data while simultaneously enabling soil moisture estimation through variations in the Received Signal Strength Indicator (RSSI), implementing the concept of joint communication and sensing (JCAS). Without deploying additional sensors, both shallow soil moisture and surface water levels can be inferred from the RSSI fluctuations. A slave BLE module (SBM) is strategically positioned at the boundary between the shallow and deep soil layers, transmitting data to a master BLE module (MBM) located above ground. Machine Learning (ML) algorithms are applied for data fusion and signal interpretation, enabling dynamic soil moisture estimation and real-time flood detection. The MBM can be mounted on drones for large-scale farms or fixed on stands for greenhouses, offering a cost-effective, low-power, and scalable solution. This JCAS framework provides continuous, real-time monitoring, significantly improving the efficiency and accuracy of soil moisture estimation and flood prediction compared to traditional methods.