Wi-Fi signals can see through walls, revealing images of still objects and even making it possible to read 3D text. The technique could have security applications, but there are also risks to people’s privacy.
Using Wi-Fi signals to snoop isn’t a new idea, but until now tools that do this have relied on movement: one recent experiment that used three transmitters and three receivers in a room could detect the shape of people’s bodies as they passed through the space by analysing the way that signals changed through time with an AI model.
Now, Yasamin Mostofi at the University of California, Santa Barbara, and her colleagues have developed a new technique called Wiffract that can decode the shape of even motionless objects from the other side of walls by looking at the unique shape of their Wi-Fi reflections.
The technique relies on the fact that when radio waves hit a sharp or tightly curved edge, they are diffracted into a very specific shape of outgoing rays known as a Keller cone. This shape can be measured and mathematically traced back to its origin, revealing the location of the point that created it.
Mostofi says that with enough Keller cones, it is possible to build up a 3D map of points along the edges of objects within a room. In tests, the team used three off-the-shelf Wi-Fi transmitters to send wireless signals into the area and recorded them using a tower of Wi-Fi receivers mounted on a remote-controlled car. The researchers used the car to move the receivers backwards and forwards, allowing them to measure reflected signals as if they had a large 2D grid receiver.
They found that they were able to create detailed images of the edge of objects in a room, even through a wall, and could read the text of a 3D sign.
Mostofi says little is needed to make the approach work except a large receiver outside a building, but thick walls can reduce the amount of detail seen. “We only rely on measuring the received signal and do not need any specialised form of transmission,” she says. “As such, as long as the signal makes it to the receiver, it can, in principle, be used.”
The concept could have a range of smart home or security applications, but there are risks as well as opportunities, says Mostofi. “As these systems get developed further in future, privacy discussions should naturally accompany them.”