Researchers at the Karlsruhe Institute of Technology (KIT) have shown that analyzing unencrypted Wi-Fi signals (beamforming feedback) in combination with machine learning models can reproduce the outlines and movements of people and even identify them with high accuracy – even if they do not have their phones with them or are not connected to the network.
KIT showed that “radio imaging” works similarly to a camera, only instead of light, radio waves are used: routers constantly exchange short feedback signals (beamforming feedback) with devices that can be read in the range. Processing such signals with a neural network makes it possible to reproduce the contours of the body, position and characteristic movements of a person – in effect turning a set of radio data into an image. According to the authors, once trained, the model is able to identify people with practical accuracy regardless of the viewing angle or gait.
The discovery has a wide range of applications, from fire and rescue telemetry to intuitive interfaces, but also great privacy risks. The researchers warn that the technology could turn any router into a covert surveillance tool – from government agencies to commercial operators or attackers, if they gain access to the relevant data and models.
The idea of using radio waves to “paint” a scene has long been explored in academic circles: there are works on localization, tracking and presence detection through walls. However, KIT has shown a step forward – not only presence detection, but also shape reproduction and recognition of people based on movement. The technology relies on unencrypted or easily accessible channel signals; Wi-Fi standards (in particular, in the work on IEEE 802.11bf) are currently discussing mechanisms that could regulate/restrict such capabilities.
The result of the study is a wake-up call for regulators, network equipment manufacturers and users. To minimize the risks, researchers and experts advise: implementing privacy-by-design mechanisms in new versions of Wi-Fi standards (IEEE 802.11bf), providing hardware manufacturers with tools to limit public access to feedback channels, and strengthening requirements for encryption and randomization of identifiers. Human rights and privacy organizations should monitor developments in this area, and legislators should promptly assess what restrictions and checks are needed to avoid mass covert surveillance.
