Innovating Surveillance with Wi-Fi Sensing

A team of researchers from the Swedish Defense Research Agency (FOI) has introduced a novel AI-powered solution for security screening. Known as WiFireNet, the system utilizes Wi-Fi signals to detect concealed handguns, providing a discreet alternative to traditional invasive security measures like metal detectors or surveillance cameras.

According to the project's documentation, the technology operates by monitoring channel state information (CSI). As Wi-Fi signals travel through an environment, they interact with objects in their path, creating specific distortions based on the item's material, orientation, and geometry. By analyzing these disruptions, the system can determine what objects are being carried.

Training and Methodology

To develop the model, the research team compiled a comprehensive dataset called the "WiFire Dataset." This involved capturing synchronized RGB video footage alongside Wi-Fi signal reflections from individuals carrying various items. The researchers noted:

«To ensure the system could specifically distinguish a handgun from everyday items of similar shape or material, the researchers trained it on recordings of a plastic handgun, a metal water bottle, a real metal handgun, and a phone.»

By training a deep learning model on these inputs, the AI learned to correlate specific visual imagery of hidden weapons with their corresponding Wi-Fi signatures.

Performance and Future Development

The current iteration of WiFireNet shows promise but faces technical challenges:

  • Detection Precision: In controlled indoor environments with up to four individuals, the system reached a 72.5% precision rate for identifying a handgun.
  • Localization: Accuracy declined to 53.8% when attempting to pinpoint the exact location of the weapon on a person.
  • Environmental Limitations: The neural network currently struggles to maintain high accuracy when transitioned to new or untested indoor layouts.

The prototype currently relies on the outdated IEEE 802.11a wireless standard. However, the researchers plan to integrate modern Wi-Fi 7 and Wi-Fi 8 standards, which offer higher-resolution data, to improve future detection capabilities. The team has made the project open source, providing access to the code and data to encourage further academic study. While this technology offers a low-cost approach to monitoring, it has sparked ongoing debate regarding the privacy implications of using omnipresent wireless networks for surveillance.