Hey, you. Yeah, I’m talking to you. Did you know that your Wi-Fi router is spying on your every move? That’s what a group of researchers from Northwestern Polytechnical University in China found.
Well, not every router. Every second, radio signals bathe your body from all directions in the form of Wi-Fi, especially if you live in a city. Virtually every home, office, and business send out these signals all the time. It’s even possible to glimpse this first hand with an app called Architecture of Radio.
When your phone, tablet or laptop connects to a router, it sends multiple requests for information. Weather forecasts, email updates, news articles, etc. In turn, the router continuously measures how its signals travel through the air, how strong they are, and if certain obstacles block them. It then adjusts itself to compensate. But this self-improvement can also track people.
When you walk through a room bathed in Wi-Fi signals, your body affects the signal. You absorb some radio waves and reflect others. The research team from Northwestern Polytechnical University found they could analyze a signal to determine precisely how a body alters it by passing through the field. Using this technique, they can “see” if someone traces letters and numbers in the air, identify a person based on their gait, and even read their lips with scary accuracy.
In a paper [PDF], created a system called FreeSense and trained it to learn peoples’ body shapes for identification. If the team tells the router that the next person walking by is one of two people, it can correctly identify the person 95% of the time. If it has to choose between six people, it identifies the right person 89% of the time.
The research team proposes a way to integrate this technology into a smart home. For example, when a person comes home, the router senses when they enter the room and communicate with other connected devices. It could tell the lights to turn on, window shades to open/close and the heater to start.
Other teams around the world have developed similar technology. A group of Australian researchers created a system called Wi-Fi ID that focuses on gait as a way to identify people. With two people, it has a 93% accuracy when told to identify one of them. Among six people the system has a 77% accuracy rate. The idea these researchers had was a way for Wi-Fi ID to identify intruders in the house and raise an alarm.
In 2013, an MIT research duo used a router to figure out how many people were in a room and identified based arm gestures. They were in a different room. Using their technology, they were able to tell how many people were in a room from behind a wooden door, from behind a 6-inch hollow wall supported by steel beams, and even an 8-inch concrete wall. They detected messages drawn in the air from five meters away in another room with 100% accuracy.
The team showed their technology, called Emerald, to President Obama during the White House Demo Day in 2015. They want to market it towards the elderly. The device could detect activity such as falls throughout the entire home. By monitoring a person’s movements, they even hope to predict falls before they even happen.
Next, a different system called WiKey could tell which keys a person was pressing. It monitored finger movements on a keyboard. After sufficient training, the system could recognize a sentence being typed in real time with 93.5% accuracy. This research team used an off-the-shelf router with just a bit of custom programming.
The lead researcher behind WiKey, Kamran Ali, was asked whether the technology could be used to steal sensitive data. He said it only works in a controlled environment with a lot of training. “So, it is not a big privacy concern, for now, no worries there.”
Finally, a research team led by a Berkeley Ph.D. student demonstrated a technology in 2014 that was able to “hear” what people were saying. They analyzed the distortions and reflections of Wi-Fi signals given off by their mouths. Their system detected which words were spoken from a list of easy-to-read vocabulary with 91% accuracy when only one person was talking. When three people spoke at once, it achieved 74%. Let this article sink in for a few minutes and then tell us what you think below.