Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Through-Wall Imaging based on WiFi Channel State Information

Julian Strohmayer, Rafael Sterzinger, Christian Stippel, Martin Kampel

TL;DR

This work addresses through-wall imaging by converting WiFi CSI into visual representations using a multimodal MoPoE-VAE. The method jointly learns latent representations for WiFi CSI and images, employing aggregation strategies and temporal encoding to improve reconstruction fidelity. Quantitative and qualitative evaluations show that concatenation with temporal encoding (C+T) yields the best image reconstructions, suggesting practical viability for privacy-preserving, camera-free monitoring and image-based downstream tasks. Overall, the approach enhances the interpretability of WiFi CSI and enables low-cost, through-wall visual sensing with potential real-world applications.

Abstract

This work presents a seminal approach for synthesizing images from WiFi Channel State Information (CSI) in through-wall scenarios. Leveraging the strengths of WiFi, such as cost-effectiveness, illumination invariance, and wall-penetrating capabilities, our approach enables visual monitoring of indoor environments beyond room boundaries and without the need for cameras. More generally, it improves the interpretability of WiFi CSI by unlocking the option to perform image-based downstream tasks, e.g., visual activity recognition. In order to achieve this crossmodal translation from WiFi CSI to images, we rely on a multimodal Variational Autoencoder (VAE) adapted to our problem specifics. We extensively evaluate our proposed methodology through an ablation study on architecture configuration and a quantitative/qualitative assessment of reconstructed images. Our results demonstrate the viability of our method and highlight its potential for practical applications.

Through-Wall Imaging based on WiFi Channel State Information

TL;DR

This work addresses through-wall imaging by converting WiFi CSI into visual representations using a multimodal MoPoE-VAE. The method jointly learns latent representations for WiFi CSI and images, employing aggregation strategies and temporal encoding to improve reconstruction fidelity. Quantitative and qualitative evaluations show that concatenation with temporal encoding (C+T) yields the best image reconstructions, suggesting practical viability for privacy-preserving, camera-free monitoring and image-based downstream tasks. Overall, the approach enhances the interpretability of WiFi CSI and enables low-cost, through-wall visual sensing with potential real-world applications.

Abstract

This work presents a seminal approach for synthesizing images from WiFi Channel State Information (CSI) in through-wall scenarios. Leveraging the strengths of WiFi, such as cost-effectiveness, illumination invariance, and wall-penetrating capabilities, our approach enables visual monitoring of indoor environments beyond room boundaries and without the need for cameras. More generally, it improves the interpretability of WiFi CSI by unlocking the option to perform image-based downstream tasks, e.g., visual activity recognition. In order to achieve this crossmodal translation from WiFi CSI to images, we rely on a multimodal Variational Autoencoder (VAE) adapted to our problem specifics. We extensively evaluate our proposed methodology through an ablation study on architecture configuration and a quantitative/qualitative assessment of reconstructed images. Our results demonstrate the viability of our method and highlight its potential for practical applications.
Paper Structure (14 sections, 4 equations, 6 figures, 1 table)

This paper contains 14 sections, 4 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Data
  4. Architecture
  5. Image Reconstruction
  6. Aggregation Options
  7. Temporal Encoding
  8. Evaluation
  9. Model Training
  10. Quantitative Results
  11. Qualitative Results
  12. Limitations
  13. Conclusion
  14. Acknowledgments

Figures (6)

  • Figure 1: Example of an image reconstructed from through-wall WiFi CSI. (a) Shows the ground truth image captured by a camera within the room and (b) the corresponding image reconstruction from (c) the CSI amplitude spectrogram recorded by a WiFi antenna located outside of the room.
  • Figure 2: Overview of the experimental setup, showing the arrangement of WiFi transmitter (TX), WiFi receiver (RX), and RGB camera (C) in the office recording environment.
  • Figure 3: Example of a CSI amplitude spectrogram showing the amplitudes of 52 L-LTF subcarriers over a $\sim$1.5-second time interval and the image corresponding to the central packet.
  • Figure 4: Proposed MoPoE-VAE architecture for WiFi CSI-based image synthesis.
  • Figure 5: Comparison of reconstruction fidelity between MoPoE-VAE models employing the aggregation options (b) uniform weighing (UW), (c) Gaussian weighing (GW), (d) concatenation (C), and (e) concatenation with temporal encoding (C+T). This visual comparison highlights the improvements in image clarity and reduction of artifacts.
  • ...and 1 more figures