> Researchers in Italy have developed a way to create a biometric identifier for people based on the way the human body interferes with Wi-Fi signal propagation.. can re-identify a person in other locations most of the time when a Wi-Fi signal can be measured. Observers could therefore track a person as they pass through signals sent by different Wi-Fi networks – even if they’re not carrying a phone.. their technique makes accurate matches on the public NTU-Fi dataset up to 95.5 percent.
Off the top of my head, I bet body composition combined with gait analysis would be enough to uniquely identify an individual.
Let's say you visit a friend in a different city, the same ISP controlling their router, can use your mac, but even if you turn off your wifi or leave your phone in your car, your volume profile and gait can betray you. how you sit, how you lean, how you turn. I'd wager, if 6-10 distinct "points" can be made out and associated with a person, that's all that's needed to uniquely identify that person after enough analysis of their motion, regardless of where they go in the world.
Imagine if they're not using one AP, but using your neighbors AP as well, two neighbor APs and your own can triangulate and refine much better.
Your resolution limit is about 30mm as a result.
WiFi presence detection is a completely different problem. If the WiFi environment is changing past a threshold, return a boolean yes or no. It can't actually tell if someone is present or if the environment is just changing, such as a car driving close enough to reflect signals back in a certain way.
Doing mass surveillance where you detect individual people in a random home environment isn't the same thing at all. All of these "could" claims are trying to drawn connections between very different problems.
With gait analysis for example, it's only looking at a handful of data points, the way we walk is very unique. lip-reading, i can see how that's a stretch, but out movement patterns and gait are disturbances in radio waves. If you're using just one person's wifi, that sounds difficult, but if you're collecting signal from multiple adjacent wifi access points, it's more realistic to build a very coarse motion representation, perhaps with a resolution no finer than 1 cubic ft, but even with more coarse representations, gait can be observed.
Even gait aside, the volume profile of a person and their location in the house alone are important data points, couple that with the unique wifi identifier or IP, you can make a really good guess at who the person is, and what room they're in.