Italian researchers have developed a pioneering method that allows for the precise tracking of human movement within enclosed spaces, leveraging standard Wi-Fi signals. The comprehensive details of their innovative work have been made public on the arXiv preprint platform.
Their newly devised approach facilitates the creation of a distinct biometric profile for individuals. This technique operates on the principle that a human body inherently modifies the shape and propagation patterns of Wi-Fi signals. By employing a sophisticated neural network, these subtle signal alterations can be meticulously analyzed, thereby enabling the continuous monitoring of people`s presence and trajectories.
The creators of this method, dubbed “WhoFi,” assert that it offers significant advantages over conventional video surveillance systems. Notably, the WhoFi system is fully operational in complete darkness and possesses the capability to penetrate through walls, all while remaining entirely imperceptible to those being monitored.
Elaborating on their findings, the researchers explained: “The fundamental concept behind our study is that as a Wi-Fi signal traverses an environment, its characteristics change based on the existence and physical properties of objects, including individuals, situated along its path.” They further state that the data derived from these changes can be utilized to construct a unique biometric “signature” for each person.
The scientists also drew a comparison to the EyeFi technology, developed in 2020, which aimed for similar tracking functionalities. However, EyeFi achieved an accuracy rate of 75%, whereas the WhoFi method demonstrates a remarkable improvement, boasting an accuracy of up to 95.5%.
In a related context, it`s worth recalling that analyst Nikita Novikov previously identified smartphone overheating as a potential indicator of unauthorized tracking. He clarified that such overheating is frequently observed when malicious software is actively running on a device.
Rupert Blackwood 
Barnaby Whitfield