Researchers at La Sapienza University of Rome have developed a novel surveillance technology that can identify individuals with high accuracy based solely on the way their bodies affect Wi-Fi signals. The system, called WhoFi, uses standard wireless networks to recognize people through passive signal analysis, without requiring any device or active participation from the subject.
Biometric Signatures from Wi-Fi Signal Distortions
WhoFi operates by analyzing Wi-Fi Channel State Information (CSI), capturing the unique distortions created as signals bounce off physical objects in an environment. Every human body affects Wi-Fi signals differently based on its shape, size, and composition. These differences generate distinctive interference patterns that serve as biometric identifiers.
The researchers explain that WhoFi focuses on multipath propagation – how Wi-Fi signals reflect off surfaces like walls, furniture, and human bodies. By collecting and processing CSI data, the system can extract a consistent pattern, or “signature,” that can be linked to an individual.
The system was trained on data collected from 30 volunteers in office and hallway settings. Using neural networks and machine learning algorithms, WhoFi learned to distinguish between individuals based on their unique signal distortion patterns. In controlled environments, it achieved a recognition accuracy of approximately 93 percent. On benchmark datasets like NTU-Fi, performance reached up to 95.5 percent using advanced transformer models.
Passive Tracking Raises Privacy Concerns
Unlike conventional biometric systems that rely on dedicated cameras or fingerprint scanners, WhoFi functions entirely passively. It does not require the subject to carry a device or interact with any hardware, raising concerns about its potential use in covert surveillance. Because it can track people without their knowledge or consent, privacy advocates have called for the establishment of ethical and legal frameworks to govern its deployment.
The researchers acknowledge that current results are based on relatively small sample sizes and controlled conditions. It remains uncertain how the system would perform in more complex environments with larger populations or frequent movement.
Despite these limitations, the implications are far-reaching. WhoFi could be used in applications ranging from security and smart building management to personalized retail environments. But without clear regulations, its capacity for passive identification has sparked alarm among civil liberties groups.
“The core insight is that Wi-Fi signals contain rich biometric information that can be decoded using deep learning methods,” the researchers note in their paper. The finding underscores the growing role of ambient sensing technologies in biometric identification—advancing surveillance capabilities while complicating long-standing privacy norms.
Sources: Interesting Engineering, arXiv
–
By Cass Kennedy
