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.

What problem does this paper attempt to address?

This paper attempts to address the problem of achieving visual monitoring of indoor environments through WiFi Channel State Information (CSI) synthesis in the context of wall penetration. Specifically, the authors propose a novel method that generates images using WiFi CSI without the need for cameras, thereby enhancing the interpretability of WiFi CSI and unlocking image-based downstream tasks such as visual activity recognition. ### Main Issues: 1. **Interpretability of WiFi CSI**: Traditional WiFi signals are primarily used for communication, but directly inferring human activities or behaviors from WiFi CSI is very difficult, if not impossible. 2. **Visual Privacy Protection**: In applications where privacy needs to be protected, using traditional cameras for monitoring poses a risk of privacy leakage. 3. **Visual Monitoring Through Walls**: Existing WiFi CSI applications are mostly limited to local indoor monitoring and cannot achieve visual monitoring through walls. ### Solutions: - **Multimodal Variational Autoencoder (MoPoE-V AE)**: The authors use a multimodal variational autoencoder to generate images from WiFi CSI, achieving visual monitoring through walls. - **Dataset Construction**: The authors constructed a dataset containing WiFi CSI and corresponding images for training and evaluating the model. - **Architecture Optimization**: The model's performance was optimized through different aggregation options (such as uniform weighting, Gaussian weighting, concatenation, etc.) and temporal encoding techniques. ### Experimental Results: - **Quantitative Evaluation**: The performance of different model variants was evaluated using metrics such as Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Root Mean Square Error (RMSE), and Fréchet Inception Distance (FID). - **Qualitative Evaluation**: Visual comparisons were made to show the impact of different aggregation options on image reconstruction quality. ### Conclusion: This method successfully synthesized images from WiFi CSI, demonstrating its feasibility in visual monitoring through walls and enhancing the interpretability of WiFi CSI. This provides a foundation for further research and applications in the future.

Through-Wall Imaging based on WiFi Channel State Information

Through-the-Wall Imaging Using Wi-Fi Signals

Vision Reimagined: AI-Powered Breakthroughs in WiFi Indoor Imaging

Building Layout Reconstruction in Concealed Human Target Sensing via UWB MIMO Through-Wall Imaging Radar.

Through-the-Wall Imaging Exploiting 2.4GHz Commodity Wi-Fi.

Two-dimensional through-wall imaging based on nonlinear inversion algorithms

Single Channel Frequency-Diverse Computational Imaging System for Through-the-Wall Sensing

Multi-View Millimeter-Wave Imaging over Wireless Cellular Network

3D Environment Sensing with Channel State Information Based on Computational Imaging

Wi-fi see it all: generative adversarial network-augmented versatile wi-fi imaging

Vision-Aided Frame-Capture-Based CSI Recomposition for WiFi Sensing: A Multimodal Approach

Experimental Investigation on Through-Wall Imaging Based on Non-Linear Inversions

We Can "See" You via Wi-Fi - WiFi Action Recognition via Vision-based Methods

High-Resolution WiFi Imaging with Reconfigurable Intelligent Surfaces

Convolutional Neural Networks Based Indoor Wi-Fi Localization with a Novel Kind of CSI Images

Through-Wall Human Pose Estimation Using Radio Signals

Directional Antenna Systems for Long-Range Through-Wall Human Activity Recognition

Through-Wall Human Pose Reconstruction via UWB MIMO Radar and 3D CNN

A Software-Defined and Distributed Wi-Fi Channel-State Information Acquisition Testbed

Augmenting visual SLAM with Wi-Fi sensing for indoor applications

A Novel Autofocusing Approach for Real-Time Through-Wall Imaging under Unknown Wall Characteristics.