What it does
Wi-Fi Guardian uses existing Wi-Fi signals to detect movement, falls, and breathing anomalies without cameras or wearables, providing private monitoring for elderly care and smart homes.
Your inspiration
We wanted to protect vulnerable individuals, such as the elderly, who are at a high risk of falls but often refuse to wear emergency pendants or feel uncomfortable with cameras in private spaces like bathrooms. The inspiration came from realizing that Wi-Fi waves already blanket our homes and constantly reflect off our bodies; the infrastructure for passive, invisible sensing was already there waiting to be utilized.
How it works
The system leverages the existing Wi-Fi router in a home as the primary signal transmitter. As standard Wi-Fi waves from the router bounce off walls, objects, and people, they create a specific electromagnetic pattern in the room. A single, low-cost microcontroller acting as a receiver captures this invisible pattern—specifically the Channel State Information (CSI). When a person moves or falls, it alters the signal reflections. We collect this raw data and use mathematical algorithms (Singular Value Decomposition) to filter out static environmental noise. By calculating the variance of the clean signal, a sudden spike alerts us to movement, while a sharp drop indicates a potential fall.
Design process
We started by researching Wi-Fi CSI capabilities using standard ESP32 development boards. Our first major roadblock was the massive amount of signal noise caused by static objects in a room. We solved this by writing a Python script that implements Singular Value Decomposition (SVD) to mathematically strip away the static background data, isolating only the dynamic signals of human movement. We then built and tested a live prototype that successfully detects movement across a room in real-time by analyzing this signal variance.
How it is different
Unlike traditional security or fall-detection systems, Wi-Fi Guardian requires zero wearables, zero cameras, and zero blind spots within the Wi-Fi coverage area. It is entirely passive and completely preserves user privacy because it only tracks "electromagnetic signal disruptions," not optical images. This makes it safe and acceptable for deployment in private areas like bedrooms and bathrooms where traditional cameras are strictly prohibited.
Future plans
Our next technical step is integrating Machine Learning models (like Convolutional Neural Networks) directly into our processing pipeline to classify specific human actions—such as distinguishing a dangerous fall from simply sitting down on a couch. Commercially, we aim to package the system as a plug-and-play smart home integration for eldercare facilities and private homes, offering peace of mind without compromising dignity
Share this page on