Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

VibraWave: Sensing the Pulse of Polluted Waters

Sagnik Ghosh, Sandip Chakraborty

TL;DR

VibraWave addresses real-time, non-invasive sensing of multiple water pollutants using a novel combination of mmWave radar and controlled acoustic excitation. The method builds a phase–AoA–range tensor across acoustic tones, applies rank-$k$ PARAFAC decomposition to extract pollutant fingerprints, and uses NNLS unmixing followed by a knowledge-distilled ResMLP for joint classification and concentration estimation. Key contributions include a physics-informed multilayer dielectric model, a 3D tensor sensing framework with fingerprint dictionaries of size $D=135$, and a distillation-based lightweight inference pipeline that achieves an overall classification accuracy of about $0.85$ and per-component RMSE around $0.20$, with robustness to tilt, placement, and reflector variations. The approach offers a portable, scalable alternative to lab assays for real-time water quality monitoring, capable of handling pure, binary, and tertiary mixtures of heavy metals, oil emulsions, and sediments.

Abstract

Conventional methods for water pollutant detection, such as chemical assays and optical spectroscopy, are often invasive, expensive, and unsuitable for real-time, portable monitoring. In this paper, we introduce VibraWave, a novel non-invasive sensing framework that combines mmWave radar with controlled acoustic excitation, tensor decomposition, and deep learning to detect and quantify a wide range of water pollutants. By capturing radar reflections as a three-dimensional tensor encoding phase dynamics, range bin power, and angle-of-arrival (AoA), we apply PARAFAC decomposition with non-negative constraints to extract compact, interpretable pollutant fingerprints. These are used to train a lightweight student neural network via knowledge distillation, enabling joint classification and quantification of heavy metals (Cu, Fe, Mg), oil emulsions, and sediments. Extensive experiments show that VibraWave achieves high accuracy and low RMSE across pure, binary, and tertiary mixtures, while remaining robust and computationally efficient, making it well-suited for scalable, real-time water quality monitoring.

VibraWave: Sensing the Pulse of Polluted Waters

TL;DR

VibraWave addresses real-time, non-invasive sensing of multiple water pollutants using a novel combination of mmWave radar and controlled acoustic excitation. The method builds a phase–AoA–range tensor across acoustic tones, applies rank- PARAFAC decomposition to extract pollutant fingerprints, and uses NNLS unmixing followed by a knowledge-distilled ResMLP for joint classification and concentration estimation. Key contributions include a physics-informed multilayer dielectric model, a 3D tensor sensing framework with fingerprint dictionaries of size , and a distillation-based lightweight inference pipeline that achieves an overall classification accuracy of about and per-component RMSE around , with robustness to tilt, placement, and reflector variations. The approach offers a portable, scalable alternative to lab assays for real-time water quality monitoring, capable of handling pure, binary, and tertiary mixtures of heavy metals, oil emulsions, and sediments.

Abstract

Conventional methods for water pollutant detection, such as chemical assays and optical spectroscopy, are often invasive, expensive, and unsuitable for real-time, portable monitoring. In this paper, we introduce VibraWave, a novel non-invasive sensing framework that combines mmWave radar with controlled acoustic excitation, tensor decomposition, and deep learning to detect and quantify a wide range of water pollutants. By capturing radar reflections as a three-dimensional tensor encoding phase dynamics, range bin power, and angle-of-arrival (AoA), we apply PARAFAC decomposition with non-negative constraints to extract compact, interpretable pollutant fingerprints. These are used to train a lightweight student neural network via knowledge distillation, enabling joint classification and quantification of heavy metals (Cu, Fe, Mg), oil emulsions, and sediments. Extensive experiments show that VibraWave achieves high accuracy and low RMSE across pure, binary, and tertiary mixtures, while remaining robust and computationally efficient, making it well-suited for scalable, real-time water quality monitoring.

Paper Structure

This paper contains 44 sections, 27 equations, 16 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminaries and Core Idea
  4. Basics of mmWave Sensing
  5. Core Idea
  6. Electromagnetic Reflections in Vibrated Liquids
  7. Multilayer Dielectric Model for Signal Reflection
  8. Analytical Derivation of the Reflection Coefficient
  9. Reflection Contributions:
  10. Special Case: Symmetric Environment ($Z_1 = Z_3$)
  11. Physical Interpretation: Signal Strength and Loss
  12. Phase Shift Induced by Acoustic Vibration
  13. Pilot Experimental Validation
  14. Power-Based Material Differentiation
  15. Phase and Frequency Response to Concentration Variations
  16. ...and 29 more sections

Figures (16)

  • Figure 1: Block diagram of VibraWave.
  • Figure 2: Electromagnetic wave interaction at an air–liquid–substrate interface.
  • Figure 4: System block diagram for end-to-end pollutant detection using VibraWave.
  • Figure 5: ResMLP student network for pollutant classification and mixture regression.
  • Figure 6: Data Acquisition Setup
  • ...and 11 more figures