A revolutionary technique presented by researchers at the University of California at Santa Barbara in the USA proposes to create images of static objects using the signal from common Wi-Fi transceivers available on the market, and to do so through solid walls. Although the ability of Wi-Fi to detect movement through walls has been known since 2015, the technology cannot “see” static objects.
The new method uses the principles of the so-called Geometric Theory of Diffraction (GTD), a technique created in 1962 by American mathematician Joseph Keller to analyze the scattering and radiation of electromagnetic waves in objects. The first task, successfully accomplished, was to read the English alphabet through walls using Wi-Fi.
To literally circumvent the absence of movement in the objects observed, the authors focused on their edges, as anyone would outline a drawing. In the experiment, the Wi-Fi signals created “Keller cones”, shapes that diffract the edges of objects for later interpretation and gradual revelation of the observed scene.
How did scientists use WiFi to read through walls?
The “Wiffract” setup, put together by researchers at UC Santa Barbara’s Mostofi Lab, consists of three WiFi transmitters and a mobile receiver to send and receive signals through walls. Although going through walls is a known feature of any home router, the issue here was to read the way in which these radio waves are affected when they hit objects.
“We then developed a mathematical framework that uses these conical footprints as signatures to infer the orientation of edges, thus creating an edge map of the scene,” explains electrical and computer engineer Yasamin Mostofi, a UCSB professor and lead author of the paper “Keller Cone Analysis for RF Images”, in a release.
The result was an image projection kernel based on the proverbial cones, to guide the formation of edges. This function is fundamental “for inferring the existence/orientation of edges through hypothesis testing on a small set of possible edge orientations”, says a press release.
Refining Wi-Fi observations
The technique works as soon as the signals detect the existence of an edge. From there, the orientation of this line of encounter between two edges is chosen, based on the best match to the signature of the Keller cone, to determine a point at which observers are interested in generating images.
“Thus, once we have found the high-confidence edge points using the proposed image kernel, we propagate their information to the rest of the points using Bayesian information propagation [a statistical process],” explains first author Anurag Pallaprolu.
Not yet peer-reviewed, the study has many real-world applications, from rescuing people in disasters to “observing” the interior of rooms in smart home monitoring.
Do you have any questions? Tell us on our social networks and take the opportunity to share the article with your friends.