I’m interested in ways to make a room respond to activity without asking everyone to carry a device or stand in front of a camera. Wi-Fi sensing is one option I’ve been exploring.
A radio signal reaches a receiver by several routes. Some of it travels directly; some reflects off walls, furniture, and people. When someone moves, those reflected paths change. With the right radio and firmware, you can measure changes in how the signal arrives.
What the radio measures
The useful measurement is called channel-state information, or CSI. It describes aspects of the signal at different frequencies, including its strength and phase. Looking at those measurements over time can reveal movement between and around the radios.
This isn’t a feature you can assume every router exposes. Getting usable CSI requires compatible hardware and software. ESP32-based setups are one relatively inexpensive way to experiment with it.
The first question I’d ask of a setup is fairly modest: can it reliably distinguish a quiet room from one with movement? That’s already useful for experiments with lighting or other room controls.
What I would test
I’d start by recording the room with nobody moving, then repeat a few simple movements in consistent locations. Changing one thing at a time makes it easier to tell whether a change in the data is related to the movement.
Placement matters. So does what else is changing in the space. I’d repeat the test after moving furniture, opening a door, or changing the position of a radio. A setup that only works in one carefully arranged room needs more work before it can be relied on.
I’d also test what happens when someone stops moving. Detecting motion and detecting a person who is sitting still are different problems. That distinction matters if the result will be used to turn equipment off.
Choose the job carefully
A coarse activity signal could be enough to vary a lighting effect or help decide when a display should wake up. It isn’t automatically a reliable people counter, a pose tracker, or proof that a room is empty.
Wi-Fi sensing doesn’t produce ordinary camera footage, but that alone doesn’t settle every privacy question. What you infer, store, and share still matters. For a simple room-activity experiment, I’d keep the recorded data limited to what the experiment actually needs.
The appeal for me is that the sensing can be unobtrusive. The work is finding out whether it’s reliable enough for a particular use, then being clear about the situations where it isn’t.