Frequency-Sensitive Soil Moisture Profiling Using WiFi Sensing

Abstract

In recent times, WiFi signals have been widely used in wireless sensing applications for detecting large-scale environmental or physical characteristics, like human count and ambient temperature. Soil moisture detection using WiFi sensing is gaining interest, but the key underlying challenge is to realise contact-free sensing technology that can characterise the impact of water spreading into the soil on the Channel State Information (CSI). Therefore, we develop a framework for sensing soil water levels using CSI- based sensing. We investigate the WiFi CSI signatures pertinent to the soil water infiltration, enabling applications including soil health monitoring. In our experimental study, we use the 5GHz WiFi spectrum to implement our novel frequency-selective CSI sensing framework for soil moisture profiling using commodity Raspberry Pi devices. The experimental results verified that specific WiFi Orthogonal Frequency Division Multiplexing (OFDM) subcarriers are more sensitive to the changes in soil moisture, leading to a frequency-sensitive behaviour on CSI. Our framework exploits the changes in the pressure levels due to the water movement or varying humidity levels in the soil channel between the WiFi transmitter and receiver that leave impressions on the underlying CSI. Furthermore, this paper demonstrates a novel approach by inspecting the CSI amplitude pattern of water entering the soil at a finer level. Lastly, in contrast to the existing works, our low-cost and contact-free method for detecting soil moisture detection has been empirically shown to efficiently utilise the newly-identified frequency-selective CSI signatures for the water infiltration and humidity levels in soil channel for accurate soil moisture prediction.