Standard WiFi Signals Can Render Near-Perfect Person Identification, Study Finds

Researchers demonstrated that ordinary WiFi transmissions can be processed to create radio-based images of people and recognize individuals with almost 100% accuracy. The method works without cameras or the need for a device on the person being observed.

A study by scientists at the Karlsruhe Institute of Technology showed that standard WiFi communications can be exploited to produce radio-frequency images of people and identify them with almost 100% accuracy. By analyzing unencrypted beamforming feedback information that devices routinely send to routers, the system creates multi-angle visualizations and matches them against a trained machine-learning model within seconds.

In experiments involving 197 volunteers, the approach succeeded regardless of the subjects' walking style or viewing perspective. The researchers note that because the method does not require any special hardware, any ordinary WiFi router in homes, offices, cafés or public venues could act as a hidden surveillance device. They stress that this capability raises significant privacy concerns, particularly for use by authorities in repressive regimes, and urge the inclusion of protective measures in upcoming WiFi standards such as IEEE 802.11bf. The work was funded by the Helmholtz Engineering Secure Systems program and presented at the ACM Conference on Computer and Communications Security in Taipei.