📶 This system uses Wi-Fi waves to identify us

Every movement of our body imperceptibly alters the electromagnetic environment around us. These alterations, invisible to the naked eye, could now serve as a unique fingerprint to recognize us.

Italian researchers have developed a system capable of identifying a person solely through the disturbances they cause in a Wi-Fi network. Named WhoFi, this device requires neither a camera, nor a phone, nor a dedicated sensor. It relies on a simple observation: our bodies interact with electromagnetic waves in a way that seems unique to each individual, thus producing a signature that can be analyzed and memorized.

Wi-Fi transformed into an identification tool

The team from Sapienza University of Rome adapted ordinary routers to capture the subtle variations of a Wi-Fi signal as it passes through a space occupied by individuals. These variations are grouped under the term Channel State Information (CSI), a set of data that describes how the wave deforms.

This information is then processed by a machine learning algorithm, based on a Transformer-type architecture. The program learns to recognize and distinguish people through the specific effects they have on the waves. Each profile then becomes a kind of invisible bodily fingerprint.

During trials, conducted on a small group of volunteers and validated using a public dataset, the system achieved an accuracy rate of 95%. This result exceeds the performance of previous experiments based on the same approach.

An alternative to conventional cameras

The method differs from usual visual tools. Unlike video surveillance, it is not influenced by lighting, physical obstacles, or changes in appearance. Signals pass through walls and partitions, making discreet observation possible in cluttered environments.

This capability extends the range of action compared to fixed cameras. A single router can cover an entire space and recognize individuals moving from room to room. Moreover, no active support like a phone or magnetic card is required.

The researchers emphasize that their approach remains experimental and is currently a preliminary scientific publication. No commercial or institutional project has been announced, but the potential seems evident.