Water Level Sensing via Communication Signals in a Bi-Static System

TL;DR

This paper introduces PMNs-WaterSense, a novel water level sensing method using existing mobile network signals and advanced signal processing to achieve high accuracy without dedicated sensors.

Contribution

It presents a new scheme leveraging CSI from mobile networks, including phase offset elimination, multi-domain filtering, and Kalman unwrapping, for precise water level estimation.

Findings

Achieves 0.025 cm error with mmWave signals

Attains 0.198 cm error with LTE signals in controlled tests

Real-world river monitoring with 4.8 cm error for 1-meter change

Abstract

Accurate water level sensing is essential for flood monitoring, agricultural irrigation, and water resource optimization. Traditional methods require dedicated sensor deployments, leading to high installation costs, vulnerability to interference, and limited resolution. This work proposes PMNs-WaterSense, a novel scheme leveraging Channel State Information (CSI) from existing mobile networks for water level sensing. Our scheme begins with a CSI-power method to eliminate phase offsets caused by clock asynchrony in bi-static systems. We then apply multi-domain filtering across the time (Doppler), frequency (delay), and spatial (Angle-of-Arrival, AoA) domains to extract phase features that finely capture variations in path length over water. To resolve the $2\pi$ phase ambiguity, we introduce a Kalman filter-based unwrapping technique. Additionally, we exploit transceiver geometry to convert path length variations into water level height changes, even with limited antenna configurations. We validate our framework through controlled experiments with 28 GHz mmWave and 3.1 GHz LTE signals in real time, achieving average height estimation errors of 0.025 cm and 0.198 cm, respectively. Moreover, real-world river monitoring with 2.6 GHz LTE signals achieves an average error of 4.8 cm for a 1-meter water level change, demonstrating its effectiveness in practical deployments.