I’m Sensing a Disturbance in the Wi-Fi

Do you always feel like somebody’s watching you and you have no privacy? Well let me tell you, it is not just a dream. Between cameras and cookies, much of what we do in both the real and digital worlds is (or at least can be) tracked. That is a bit unsettling to think about, and I would like to give you some good news, but as technology advances, these snooping methods only seem to become more sophisticated. From facial recognition software to browser fingerprinting, the ways in which our activities are monitored and recorded are growing in number and complexity.

In the latest turn of events, even our Wi-Fi routers have been recruited to spy on us. A group of researchers at the University of Rome has just detailed a new surveillance method that they call WhoFi. Using a few clever tricks, they have demonstrated that it is possible to individually identify people just by analyzing the way that they disrupt the Wi-Fi signals that are around them.

The system makes use of a component of Wi-Fi signals known as Channel State Information. It captures the subtle changes in a wireless signal as it bounces around people and objects in an environment. These changes, the team found, are as unique as a fingerprint, which makes them useful for Person Re-Identification (Re-ID) applications.

Re-ID is the ability to determine whether two instances of sensor data belong to the same individual. Traditional Re-ID methods depend on visual data like body shape or clothing color, but these can be affected by lighting, occlusion, or camera placement. WhoFi sidesteps those issues entirely, opening the door for more reliable, less visible forms of tracking.

Unlike facial recognition systems or fingerprint scanners, WhoFi does not require cameras or direct physical contact, rendering it difficult to detect. It works by analyzing how a person’s presence alters Wi-Fi signals over time. That means it can operate without line of sight, even through walls, and in conditions where visual systems would fail, like in darkness, smoke, or fog.

To turn raw Wi-Fi signal distortions into identifiable patterns, the team trained a deep neural network capable of distinguishing individuals based on their unique impact on Wi-Fi signals. The system achieved a 95.5% accuracy level on a benchmark dataset, and it remained effective even when people moved to new surroundings.

The prospect of being identifiable and trackable without ever being seen or touching a device raises serious questions about privacy. The researchers acknowledge the potential for misuse, but stress that their system does not collect personal data or traditional identity information. Instead, it relies entirely on “non-visual biometric features” hidden in the Wi-Fi signals themselves. Of course, that may not be enough to ease everyone’s concerns about the technology.

There are no commercial or government plans to deploy it — at least not yet. But the researchers believe their work lays the foundation for future developments in wireless biometric sensing, a field that is likely to attract growing interest from both industry and security agencies.