Wi-Fi Signals Become a Pocket Radar for Spotting Movement Through Walls

Makers assembled a compact handheld unit that listens to standard wireless network signals and converts subtle shifts in those signals into a real-time radar display of nearby human activity. The entire system fits comfortably in one hand and comes together for roughly fifty dollars in parts. Movement registers even when walls or other barriers stand between the device and the person. No camera, no infrared sensor, and no dedicated motion detector appears anywhere in the build. Instead the unit taps into Channel State Information (CSI) carried by ordinary Wi-Fi traffic in most homes and offices.

CSI contains a wealth of information on signal intensity and phase across several sub-channels, acting as a unique fingerprint that is written into the air by every surface, item, and person in its vicinity. Then there’s human tissue, where the water reacts with radio waves in a way that makes it stand out from everything else in the room. Of course, motion alters this distinct pattern, allowing the device to detect it. After calibrating the device with a quick empty room scan, the firmware can set a baseline and compare new data to it. Any continuous changes that exceed a specific threshold will trigger a detection and leave a mark on the radar display. One processing core is constantly collecting data, while the other maintains the screens responsive, with no lag.

Building it all begins with an M5Stack Cardputer ADV board, which already has an ESP32-S3 microcontroller, a small screen, a keyboard, and a battery. Then you simply slap an extra display onto the board and place it in a simple 3D printed frame above the primary unit, and you’re ready to go. Dual displays are no problem, and because it reuses existing Wi-Fi networks rather than creating its own, the parts list is kept to a minimum.

When it first powers up, it scans for available networks; select one, input your password, and you’re ready to go. Connection is confined to the networks you manage, which prevents the unit from spying on other people’s traffic. Once the connection is established, the radar activates and the smaller screen displays all connection details, a scrolling graph of recent activity, and options such as a clear or presence banner, as well as the sensitivity setting. The larger screen displays the real radar image, which is a sweep line slowly traveling across a circular view with fading trails indicating where earlier movements were identified, and as the marks emerge, they grow or shrink depending on how much the signals shifted at the time.

Of course, there are keyboard shortcuts for things like triggering a fresh calibration, adjusting the detection threshold, and navigating to the settings menu. The entire machine can run totally on battery power or a USB connection when necessary, and because it just uses Wi-Fi signals that are already floating around, it can work in the dark and through most interior walls. With some furniture around, readings can be thrown off as reflections bounce off everything, causing your gadget to go berserk, but with careful positioning and tweaks, you can generally keep the false triggers under control. The problem is that the current design just provides you a general sense of movement in the area and does not provide precise direction or distance.

Because the code, build files, and enclosure designs are now public on Github, anybody can recreate or adapt the project to meet their own requirements. The same Wi-Fi signals that make life so convenient also allow you to utilize this device to determine whether anyone is in the room or if there’s any unusual activity going on someplace. As a result, it has a variety of helpful uses, such as automatic lighting control when you enter a room and simple notifications when someone does something unexpected. The only issue is attempting to completely avoid using it without blocking Wi-Fi connections.
[Source]