POSTECH Researchers Demo Ultrasonic Fingerprint Authentication Without Storing Digital Biometric Data

Researchers at Pohang University of Science and Technology (POSTECH) in South Korea have developed a biometric authentication system that verifies fingerprints through physical acoustic interactions rather than comparing digital image files, according to a report by TechXplore (https://techxplore.com/news/2026-10-fingerprint-authentication-ultrasound-biometric-digital.html).

The study, published in the journal Nature Sensors, was led by Professor Junsuk Rho across multiple POSTECH engineering departments alongside Dr. Seokho Lee and graduate researcher Beomseok Oh.

Conventional biometric security systems digitize fingerprint scans and evaluate them against stored reference files. Because biometric traits cannot be changed after a database exposure, storing digital templates creates a persistent target for attackers even when files are encrypted.

To remove the storage requirement, the researchers designed a metasurface-driven loop-diffractive neural network (loop-DNN). The engineered metasurface structure directs ultrasound waves based on the physical ridges of a presented fingerprint. When an authorized fingerprint is scanned, ultrasonic energy focuses onto a predefined registered region. An unauthorized print directs acoustic energy toward a non-registered region. Authentication requires only a direct signal intensity comparison between the two zones, bypassing image reconstruction and digital feature extraction algorithms.

The architecture controls ultrasonic wave phase, amplitude, and transmission switch states simultaneously, expanding wave-manipulation parameters beyond phase-only diffractive neural networks.

In laboratory testing using 192 validation samples from four individuals, the prototype achieved a 97.92% authentication accuracy rate. Numerical simulations using an international fingerprint recognition database demonstrated an accuracy rate of up to 99.75% for a single registered user.

Commercial deployment in consumer electronics and industrial hardware will require structural miniaturization. The current experimental setup operates at approximately 40 kilohertz (kHz) and must be adapted to the megahertz (MHz) frequencies standard in commercial sensors. The team also plans to expand the single-user prototype to support multiple registered users by pairing multi-layer metasurfaces with digital systems.