M-PPG: A Nonintrusive Authentication for Wearable Devices via PPG Sensors With Motor Excitations

With the popularity of wearable devices, verifying the user identity of wearable devices is critical for system security, especially when performing sensitive operations such as financial payments. In this article, we propose M-PPG, which uses photoplethysmography (PPG) measurements with motor vibrations to authenticate user identities. Specifically, our observations reveal a strong correlation between PPG measurements and users, such as the user’s heartbeat characteristics and wearing habits reflected by PPG measurements are implicitly associated with users’ identities, and the motor vibration provides stable excitation signals for better security performance. M-PPG requires zero user effort with a built-in sensor and actuator in most wearable devices, i.e., PPG sensor and motor. In addition, we reveal that indoor lighting strobe noise is an unavoidable noise in PPG-based authentication methods, and a noise reduction mechanism is designed to mitigate such phenomenon. Moreover, a multiple authentication scheme and an adaptive update scheme are proposed to keep a long-term authentication in M-PPG. The experimental results among 20 participants demonstrate that M-PPG achieves an average accuracy of 94.5% in daily situations. With a well-designed authentication mechanism, the pass rate of legitimate users can reach 99.9%, while the pass rate of attackers is only 0.6%.