Earphone sensors show potential for continuous biometric authentication

Researchers have developed an earphone-based biometric authentication system that identifies users through subtle heart and respiratory signals captured by wearable sensors.

No passwords, no face scans, no finger on a sensor – just put your AirPods in and walk up to the door. That is the vision behind AccLock, a new earphone-based authentication system developed by a team of researchers that identifies its wearer using unique physiological signals produced by the heart and lungs.

The system leverages earphone accelerometers to detect subtle sounds from the heart pumping blood and the respiratory cycle, known as ballistocardiographic (BCG) signals. Each person has unique BCG signals that are difficult to spoof because they originate in our own physiology.

The findings were submitted by researchers from universities in China, Australia and Denmark and published on ArXiv.

AccLock is designed for continuous, hands-free authentication. Rather than requiring any deliberate action, the system works simply by being worn — meaning a person could walk up to a secured door and have it unlocked automatically, just by having their earphones in.

“Earphone-based authentication emerges as a promising solution for continuous user authentication, because earphones remain closely coupled with the wearer during everyday use,” the paper says. “By continuously verifying whether the current wearer remains the legitimate user, such a capability protects persistent access to paired devices and services without requiring explicit user actions.”

The reliance on BCG signals also avoids many issues faced by other types of earphone-based identity authentication.

Developers have previously suggested that earphones verify users using behavioral biometrics, such as tooth clicking, mandible vibration, and walking patterns, which are recorded by a microphone or sensor. This, however, requires that the user explicitly perform actions such as clicking their teeth – which can be inconvenient if not odd.

Other earphone-based identity authentication systems focused on measuring the ultrasonic features of the ear canal using the earphone’s speaker and microphone. In this case, the technology relies on the speaker, which means that the wearer would not be able to use it.

Another proposed solution is to capture heart sounds, a feature enabled by in-ear microphones integrated into active noise cancellation (ANC). But this approach faces issues such as environmental noise and the higher cost of active noise cancellation tech, the paper notes.

The researcher tested the AccLock system on 33 participants using both AirPods and their own 3D-printed earphones. The study showed that it was fairly accurate, with average false acceptance rates (FAR) and equal error rates (EER) of 3.13 percent and 2.99 percent, respectively.

Article Topics

AccLock  |  biometric authentication  |  biometrics  |  biometrics research  |  heartbeat biometrics  |  wearables