Wi-Fi can accurately identify people, even if they aren't carrying a phone or computer

Daniel Sims

Posts: 2,468   +74
Staff
The takeaway: People often worry about being tracked through their wireless devices, especially when using public Wi-Fi networks. However, researchers have discovered multiple methods to detect and potentially track individuals via Wi-Fi, even if they are not carrying any devices, and the widespread presence of Wi-Fi networks makes these surveillance tactics potentially universal.

According to a recent study (PDF) from the Karlsruhe Institute of Technology, any Wi-Fi router that supports Wi-Fi 5 (802.11ac) or newer can be used to observe people within range. The findings raise serious privacy concerns.

The researchers introduced a new identity-inference attack called BFId, which exploits beamforming – a technique standardized with Wi-Fi 5 (802.11ac). The attack is possible with commercially available hardware and tracks people rather than the devices they carry, bypassing software-based security measures.

If multiple Wi-Fi devices can communicate with each other, beamforming signals can generate radio-based "images" from multiple angles, enabling the identification of people without cameras or other traditional methods. In the study, researchers were able to track 197 participants with nearly 100 percent accuracy, regardless of how they moved or the angle from which they were detected.

Once a machine learning model is trained, the system can identify targets within seconds. Moreover, because Wi-Fi signals are unencrypted, this information is accessible to anyone within range.

Most Wi-Fi devices currently in use likely support Wi-Fi 5 or newer, meaning BFId could potentially be deployed almost anywhere. Privacy advocates warn that governments, cybercriminals, or other malicious actors could exploit the technology to observe targets more discreetly than traditional surveillance methods.

For instance, earlier this year, a contractor for the city of New Orleans was revealed to have conducted AI-based surveillance across hundreds of devices for two years without public knowledge. A similar operation using Wi-Fi – based tracking could bypass laws against facial recognition, forcing regulations to play catch-up once again.

Other researchers have also explored tracking people through Wi-Fi signals independent of the devices they carry. Earlier this year, a study from La Sapienza University of Rome introduced WhoFi, which identifies people based on how their bodies disrupt Wi-Fi signals. Like BFId, WhoFi achieved a success rate of over 90 percent using deep learning models. Previous technologies have even recognized individuals through gestures and through walls.

Permalink to story:

 
Privacy no longer exists for real, we just pretend it does between people and small companies to keep alive the ilussion of control, but since the Edward Snowden incident with the NSA, there is simply no way a person can expect to not be completely spied on from the 1st world countries, and they look like the good guys so we can even expect worst from the others like Russia or China, which honestly feel like a disguised dictatorship.

And I'm saying this as an Argentinian, without favoritism to any side and I'm aware that the media is heavily manipulated.
 
I wonder how much trouble it would be to roll this out across public network meshes at a local government level and just start raking in the data collection money? Think they wouldn't? Police already work for the revenue.
 
Privacy no longer exists for real, we just pretend it does between people and small companies to keep alive the ilussion of control, but since the Edward Snowden incident with the NSA, there is simply no way a person can expect to not be completely spied on from the 1st world countries, and they look like the good guys so we can even expect worst from the others like Russia or China, which honestly feel like a disguised dictatorship.

And I'm saying this as an Argentinian, without favoritism to any side and I'm aware that the media is heavily manipulated.

Privacy doesn't exist if you make no effort to protect it. If you allow a dozen different tracking scripts to load in every web page you visit, if you put your real name in the data entry boxes, if you do other dumb ****, no, you won't have any privacy. But, if you make an effort you can minimize the data collected that can be aggregated and associated with your true identity. If you use Brave or Firefox, add the No-Scipt add-on and only give the scripts necessary to make the page load permission to load in the page. That's a big start. If the script doesn't load, it can't track you. Uses hosts file blocking to block thousands of tracking servers. Use alias email addresses (manyme.com is awesome). There is quite a bit you can do. It's not always convenient. But, the effort you make is equal to the privacy you take.
 
Terrifying.

But at least it needs to control the wifi devices to work.

Luckily, your average wifi access point has strong security.
But you can just bring your own router. This essentially works the same way a lightbulb works, it's just that it's radio instead.
 
Besides all the negatives, I wonder if this could be used to help fire fighters locate people. No idea if fire would interfere or not.
Probably not. Flames can create plasma that disrupts signals, while smoke can absorb and scatter radio waves. Both process would interfere with using Wifi to locate people.
 
Probably not. Flames can create plasma that disrupts signals, while smoke can absorb and scatter radio waves. Both process would interfere with using Wifi to locate people.
Besides all the negatives, I wonder if this could be used to help fire fighters locate people. No idea if fire would interfere or not.

Not to mention that unless the firefighters are bringing routers to the scene, they can't assume that there will be Wifi in a burning building.

That said, it's a great idea, and it wouldn't surprise me at all if there's purpose built tech already or in development for this exact use case. If AI is being used, they can train it to mitigate some of the interference commonly found at fire scenes.
 
My WiFi routers, Linksys Velop mesh their earliest try-band nodes, rolled out with WiFi motion detection back in 2019 I believe it was. I ran the free trial they gave users just to check it out because of how interesting it was.

Linksys advertised as a home security feature and it would send notifications if it detected a person when no one was supposed to be there.

It worked fairly well.
 
That said, it's a great idea, and it wouldn't surprise me at all if there's purpose built tech already or in development for this exact use case. If AI is being used, they can train it to mitigate some of the interference commonly found at fire scenes.

I have a large house I would use it just to find out which floor and what rooms the pets are in. It is very easy for them to become stuck in a room when a door is closed.
 
But you can just bring your own router. This essentially works the same way a lightbulb works, it's just that it's radio instead.
Yes but following people around with routers is much less efficient than accessing the millions of compromised routers already out there.

Further, at the point you have sent people out to track someone's movements there are far better technologies than wifi router beamformed tracking.
 
Well, if you're really smart anything is possible. There is technology that can read the data transmitted by by the transistors and circuits on the motherboards of personal computers and cell phones. Not at great range, but it can be done. It's also important to keep into perspective where the whole Wi-Fi can be intercepted scare-mongering notion originated. Happened back in the 2010s and it was one guy doing it in San Francisco sitting outside coffee bars. He was eventually caught and prosecuted. Today, Wi-Fi insecurity in public places continues to be a favorite subject for security firms and tech journalists. IMHO + 2 cents.
 
Here's the key sentence.

"If multiple Wi-Fi devices can communicate with each other, beamforming signals can generate radio-based "images" from multiple angles, enabling the identification of people without cameras or other traditional methods."

Angles are defined by three points, so it may require at least three Wifi devices. Then you have to install and train the AI software. That could get tricky.
 
Last edited:
Back