WiFi can be used to detect hidden guns and alert security, study shows

Excerpt: WiFi setup can distinguish who carries a hidden gun, and who has a water bottle

WiFi signals can be used to detect concealed handguns and cover some blind spots in surveillance cameras and metal detectors, a study by the Swedish Defense Research Agency (FOI) has proposed. The researchers demonstrated a prototype that located up to 90% of the items carried by up to 4 people walking in a room.

The experiment setup was quite rudimentary. Researchers placed 2 WiFi transmitters on one side of the room and 2 receivers on the other.

An operating frequency of 5.68 GHz is the same as on many home routers. A bandwidth of 20 MHz was used, which is the narrowest standard WiFi channel.

Data was collected using a modified open-source implementation of IEEE 802.11a, an older WiFi standard from 1999, and GNU Radio, free software for radio signal processing.

Up to 4 participants walked in the up to 12-square-meter area, covered by WiFi signals, and a standard video camera filmed everyone.

The core idea is that WiFi signals are distorted differently as they pass through people and objects, such as a metal handgun, a water bottle, a phone, or a plastic toy gun. The researchers trained a machine learning model to distinguish and localize individuals carrying a concealed handgun.

“Our baseline results demonstrate that WiFi signals indeed contain sufficient information to detect concealed handguns carried by humans in groups of up to four people and in the presence of other commonly carried items,” the researchers conclude in a paper called Concealed Handgun Detection using WiFi Signals.

How does it work?

The researchers utilized WiFi Channel State Information (CSI), a detailed measurement of how radio waves propagate, reflect, and change. Routers use it to optimize connections for reliable internet speeds.

CSI essentially slices the 20 MHz channel into 52 narrow, distinct frequency segments and measures how much each of them strengthens or weakens (amplitude) and how much its timing shifts (phase) 80 times per second (80Hz).

The AI model analyzing CSI data learned to recognize the handgun patterns and draw a 2D box around the person it thinks is carrying a gun.

Researchers trained their model to distinguish 4 common classes: a handgun, a plastic handgun, a metal water bottle, and a mobile phone.

The authors argue that CSI signals cover a wide area, can spot hidden items, and don’t need a security gate with a guard.

“A WiFi-based security system has the potential to be a complementary modality that is able to handle occlusions or concealment and is not hindered by strict spatial configurations,” the researchers explain.

The researchers call their deep learning model, tailored to detect concealed handguns in WiFi signals, WiFireNet.

According to the results, the system found 83-91.5% of objects people carried, depending on room size and number of people. However, it also makes many mistakes – the so-called “mean average precision” score, which punishes false positives, was around 35-50%.

The average precision for the handgun class alone ranges from about 30% to 72%.

The collected data, code, and trained model are now public for other labs to reproduce or improve.

WiFi signals are already being used to detect intruders, sense presence, or even see through walls. Unlike previous studies, the researchers demonstrated detection capabilities when multiple people move simultaneously.

“To our knowledge, no prior work has been done on WiFi-based handgun detection in dynamic multi-person environments.”