Detecting Object Outlines Behind Walls Using Wi-Fi Signal Analysis
A research team from the University of California, Santa Barbara has developed a method for detecting the outlines of stationary objects behind walls by analyzing distortions in Wi-Fi signals. The method, called Wiffract, is based on identifying changes in the signal caused by the interaction of electromagnetic waves from a Wi-Fi transmitter with the edges of objects.
The mathematical framework used by the researchers is based on the Geometric Theory of Diffraction (GTD), which describes the effects that occur when electromagnetic waves bend around obstacles. In addition to incident, refracted, and reflected waves, GTD introduces the concept of diffracted waves, which arise when a wave encounters an edge or sharp vertex on an object’s surface. If a wave hits an edge, the diffracted waves form the surface of a cone of rotation (Keller’s cone), with an opening angle equal to twice the angle between the incident wave and the tangent. If the incident wave is perpendicular to the tangent at the edge, the cone unfolds into a plane, and if it hits a sharp vertex, the diffracted waves spread evenly in all directions.
The proposed method does not require prior training of a neural network and is not limited to detecting only those objects included during machine learning. Instead, the neural network attempts to reconstruct the outlines of arbitrary objects by tracking their edges. The signal analyzer, which emulates an antenna array made up of Wi-Fi receivers, takes into account changes in signal strength at specific points on a two-dimensional plane. In the incoming signal, the neural network identifies distortions characteristic of diffracted waves that occur when a wave hits an edge, and reconstructs the spatial position of those edges.
To demonstrate the method, the researchers detected mock-ups of English alphabet letters placed behind a wall, using three standard wireless transmitters operating at Wi-Fi frequencies. For signal reception, they created a scanning cart that moved back and forth, equipped with several Wi-Fi receivers to emulate an antenna array. Notably, the method works not only for objects with visible sharp edges but is also applicable to objects with a small degree of surface curvature.