RuView: Using WiFi Signals to Track Movement Without Cameras

You know that feeling when you walk into a room and the lights turn on automatically? Usually that's a motion sensor — a little PIR device on the ceiling. But what if I told you your WiFi router could do the same thing, without any extra hardware?

That's the idea behind RuView: using the WiFi signals already bouncing around your home to detect where people are and what they're doing. No cameras, no wearables, no new devices to buy.

I've been reading about this for a while, and I think it's one of those technologies that sounds like science fiction but is actually running in people's homes right now. Let me break down how it works, what it can actually do, and where the limits are.

The Basic Idea

WiFi signals are radio waves. They travel through the air, bounce off walls, get absorbed by furniture, and — this is the key part — get disrupted by human bodies. You're mostly water, and water absorbs WiFi signals in predictable ways.

When someone walks through a room, they slightly change the pattern of WiFi signal strength in that area. These changes are tiny, but they're measurable. RuView captures these fluctuations and uses machine learning to figure out what's causing them.

Think of it like this: imagine you're in a dark room with a bunch of flashlights pointing in different directions. You can't see anything, but you can see the shadows on the walls. Now someone walks through the room — the shadows change. You don't need to see the person to know they're there. RuView does the same thing, but with radio waves instead of light.

Why This Matters for Privacy

Here's where it gets interesting. The biggest problem with smart home sensing is privacy. Cameras are the most accurate way to track movement in a home, but most people don't want cameras in their bedrooms, bathrooms, or living rooms. The psychological barrier is real.

WiFi sensing sidesteps this entirely. There's no video, no audio, no identifiable imagery. The system only sees signal strength fluctuations — numbers going up and down. It can tell someone is in the room and roughly where they are, but it can't tell who they are or what they look like.

For anyone building smart home systems that need spatial awareness without the creepiness factor, this is a big deal.

How RuView Actually Works

The technical core is surprisingly straightforward, though the implementation details get complex.

Step 1: Collect signal data. You deploy a few small devices (Raspberry Pi works) in the target area. These devices continuously monitor WiFi signal strength — specifically the RSSI (Received Signal Strength Indicator) values from all nearby WiFi access points.

Step 2: Label the data. You perform specific actions while the system records — walk from the kitchen to the living room, sit on the couch, open the fridge. Each action gets tagged with the corresponding signal data.

Step 3: Train a model. The labeled data feeds into a neural network (CNN or RNN) that learns to associate specific signal patterns with specific movements or positions.

Step 4: Deploy for real-time inference. Once trained, the model runs on your edge devices and outputs position coordinates in real time. You can pipe these into Home Assistant or any smart home platform via MQTT.

The Hard Parts

I don't want to oversell this. There are real technical challenges that make this harder than it sounds.

Multipath interference is the biggest headache. In a typical indoor environment, WiFi signals bounce off walls, furniture, appliances — you name it. The signal your receiver picks up is a jumbled mess of direct and reflected paths. This "multipath effect" adds so much noise that raw RSSI data is almost useless for precise positioning. You need sophisticated signal processing just to get a clean signal.

Signal attenuation is non-linear. In theory, WiFi signal strength decreases predictably with distance. In practice, a human body blocking the signal causes sudden drops that don't follow any simple formula. You can't just use a physics equation to calculate position — you need the neural network to learn the specific signal behavior of your specific environment.

Every space is different. A model trained in my apartment won't work in yours. The layout, the furniture, the wall materials, even the number of people living there — all of these change the signal patterns. This means you need to retrain (or at least fine-tune) the model for every new deployment.

Hardware differences matter. Different WiFi chipsets report RSSI values differently. Two devices in the same location might report signal strength that differs by 5-10dB. Without normalization, this hardware-level inconsistency can prevent the model from converging.

What It's Actually Good At (Right Now)

Given those limitations, what can you realistically do with WiFi sensing today?

Presence detection — knowing whether someone is in a room — is the most reliable use case. This is what the basic version of RuView handles well. It's not perfect, but it's good enough for automations like "turn off the lights when no one's been in the room for 10 minutes."

Rough positioning — knowing which room someone is in — is achievable with enough collection nodes. You won't get centimeter-level accuracy, but room-level tracking is realistic.

Gesture recognition — detecting specific large movements like walking, sitting, or waving — has been demonstrated in research settings. The accuracy depends heavily on training data quality.

What it can't do well yet: precise tracking (like following someone's exact path through a house), identifying individuals, detecting small movements (like typing), or working reliably in environments with lots of electromagnetic interference.

My Honest Take

WiFi sensing is genuinely promising, but it's still in the "early adopter" phase. The technology works, but it requires technical know-how to set up, and the results are inconsistent compared to cameras or dedicated sensors.

The privacy angle is the real selling point. If you're someone who wants smart home automation but refuses to put cameras in your home, WiFi sensing is one of the only options that gives you meaningful spatial awareness without compromising privacy.

I think the technology will get significantly better over the next few years as models improve and hardware gets cheaper. But right now, if you're not comfortable with Python scripts, neural network training, and debugging signal processing issues, this isn't a plug-and-play solution yet.

For developers and tinkerers, though? It's a fascinating project to experiment with. The fact that you can turn existing WiFi infrastructure into a sensor network with a few Raspberry Pis is pretty remarkable.

Just go in with realistic expectations — it's clever, it's privacy-friendly, but it's not magic.

If you are intrigued by WiFi sensing but not ready to dive into the deep end, a good intermediate step is to experiment with the RSSI data itself before attempting full localization. Simply placing a Raspberry Pi near your WiFi router and logging RSSI values over a few days will reveal just how much signal strength fluctuates with human presence, time of day, and even weather conditions. This data alone can be useful for basic presence detection without any machine learning at all — just set a threshold and trigger an automation when signal strength drops below it. It is far less accurate than the full RuView approach, but it requires almost zero setup and can be a satisfying first step into the world of invisible sensing. Looking further ahead, the convergence of WiFi sensing with other low-cost sensing technologies — such as passive infrared sensors, ultrasonic range finders, and even ambient light and sound sensors — could produce hybrid systems that are far more robust and accurate than any single technology alone. Imagine a home automation system that fuses WiFi signal disturbance patterns with ambient sound levels and temperature changes to determine not just whether someone is in a room, but what they are doing: cooking in the kitchen, watching television, or sleeping. This kind of sensor fusion is already being explored in research labs and could eventually make its way into consumer products. The most exciting implication of WiFi sensing technology for the broader smart home industry is the potential to finally achieve ubiquitous presence-aware automation without the privacy concerns that have held back camera-based systems. If the accuracy continues to improve and the setup process becomes more accessible, WiFi sensing could become a standard feature in future WiFi routers, effectively turning every home with internet connectivity into a spatially-aware environment capable of sophisticated automations that today require expensive dedicated sensor installations.

As sensing technologies mature and converge, the invisible infrastructure of smart environments will finally match the reliability and privacy standards that consumers expect.

Looking further ahead, the convergence of WiFi sensing with other low cost sensing technologies could produce hybrid systems far more robust and accurate than any single technology alone. Imagine a home automation system that fuses WiFi signal disturbance patterns with ambient sound levels and temperature changes to determine not just whether someone is in a room but what they are doing, such as cooking, watching television, or sleeping. Passive infrared sensors, ultrasonic range finders, and ambient light sensors each capture a different slice of environmental state. When a central hub combines these signals, the resulting picture of home occupancy becomes remarkably reliable. The key enabler is cheap edge compute: modern microcontrollers cost just a few dollars and run lightweight machine learning models efficiently. As these technologies mature, invisible sensing infrastructure will finally match the reliability and privacy standards that consumers expect from smart home systems.