FOI researchers test Wi-Fi system for concealed handgun detection

The prototype points to a less intrusive screening method, but its current accuracy is uneven in controlled indoor tests.

Researchers at Sweden’s FOI have built an experimental AI system, WiFireNet, that tries to spot concealed handguns by analyzing how Wi‑Fi signals are distorted, which could offer a less intrusive screening method than cameras or metal detectors. The team trained the model on a dataset built from paired video and Wi‑Fi recordings of people carrying items including phones, plastic and metal handguns, and a metal water bottle. In controlled indoor tests with up to four moving people, the system reached 72.5% average precision for handgun detection, though accuracy fell to 53.8% when the goal was to locate the weapon exactly on the carrier. The prototype used modified IEEE 802.11a hardware, and the researchers say later versions should use Wi‑Fi 7 and Wi‑Fi 8 for richer signal data. They also released the code, models, and dataset as open source.