Researchers train AI to detect concealed handguns using Wi-Fi signals

Security researchers have trained an artificial intelligence system to identify concealed handguns by analyzing the way ordinary Wi-Fi signals reflect off a person and the objects they carry. The experimental system, called WiFireNet, detects weapons solely from Wi-Fi channel state information, the data produced when signals pass through and bounce off objects in ways shaped by their geometry, orientation, and material.

The work builds on earlier security research in which ordinary Wi-Fi routers were shown to identify people with 99.5% accuracy. To develop WiFireNet, the researchers assembled the WiFire Dataset by recording RGB video and reflected Wi-Fi signals at the same time as people walked around carrying various objects, from phones to handguns. They then trained a deep learning model on the dataset, using the video clips to pair the visual appearance of a concealed handgun with its corresponding Wi-Fi signature.

To make sure the system could tell a handgun apart from everyday items of similar shape or material, the team trained it on recordings of a plastic handgun, a metal water bottle, a real metal handgun, and a phone. In controlled trials that tracked up to four moving individuals in a small indoor space, WiFireNet achieved an average handgun-detection precision of 72.5%. When the system was held to stricter localization requirements, such as pinpointing the weapon's exact location on the carrier, precision fell to 53.8%. The neural network also performs poorly in unfamiliar indoor layouts it was not tested on.

The prototype was built on a modified implementation of the older 1999 IEEE 802.11a wireless standard to establish proof of concept. Planned later phases call for modern Wi-Fi 7 and Wi-Fi 8 networks, whose higher-resolution channel state information is expected to improve detection. The researchers released the project's code, models, and dataset as open source to support further research.

Wi-Fi sensing is drawing growing interest as a surveillance technology because it does not require complex equipment or setups, and critics contend that the technology's ubiquity, combined with a lack of regulatory frameworks, raises serious privacy concerns. The system remains experimental, and the researchers have outlined a roadmap of planned improvements.

More AI news from TechManNews.