Ear Acoustic Biometrics Using Inaudible Signals and Its Application to Continuous User Authentication
Shivangi Mahto, Takayuki Arakawa, Takafumi Koshinak · 2018
This paper presents an improved version of a previously-proposed ear-acoustic biometric system for personal authentication. Even though the previous system provided a fast, accurate, and easy means of authentication, it employed noticeably audible probe signals to extract ear acoustic-features, signals which might interrupt user activities. To overcome this problem, this paper presents silent user authentication by employing inaudible signals in the place of audible signals for capturing ear acoustic-features. A comparative study using a number of audible and inaudible signals demonstrates that inaudible signals provide accurate authentication under the condition that the relative position of the earphone device against the ear canal is constant, which is a requirement for continuous user authentication. On the other hand, audible signals offer better accuracy when the earphone position changes, which is often the case in initial user authentication. This suggests the idea of a hybrid system that employs both audible and inaudible signals for, respectively, accurate initial authentication and user-friendly continuous authentication.