TY - JOUR
T1 - HeadSonic
T2 - Usable Bone Conduction Earphone Authentication via Head-Conducted Sounds
AU - He, Zhixiang
AU - Chen, Jing
AU - He, Kun
AU - Gu, Yangyang
AU - Deng, Qiyi
AU - Zhang, Zijian
AU - Du, Ruiying
AU - Zhao, Qingchuan
AU - Wu, Cong
N1 - Publisher Copyright:
© 2002-2012 IEEE.
PY - 2025
Y1 - 2025
N2 - Earables (ear wearables) are rapidly emerging as a new platform encompassing a diverse of personal applications, prompting the development of authentication schemes to protect user privacy. Existing earable authentication methods are all specifically designed for air-conduction earphones, which are not suited for bone conduction earphones (BCEs) that rely on bone conduction mechanisms. In this paper, we propose HeadSonic, a usable BCE authentication system based on the unique head-conducted sounds, which can be acquired when the user wears the BCE device. Specifically, the system emits a millisecond-level sound to initiate the authentication session. The signal captured by the BCE microphone is propagated through the user's head, which is unique in density, geometry, and bone-tissue ratio. It operates implicitly, while maintaining robustness across different behaviors. Extensive experiments involving 60 subjects demonstrate that HeadSonic achieves a commendable balanced accuracy of 96.59%, proving its efficacy and resilience against replay and synthesis attacks.
AB - Earables (ear wearables) are rapidly emerging as a new platform encompassing a diverse of personal applications, prompting the development of authentication schemes to protect user privacy. Existing earable authentication methods are all specifically designed for air-conduction earphones, which are not suited for bone conduction earphones (BCEs) that rely on bone conduction mechanisms. In this paper, we propose HeadSonic, a usable BCE authentication system based on the unique head-conducted sounds, which can be acquired when the user wears the BCE device. Specifically, the system emits a millisecond-level sound to initiate the authentication session. The signal captured by the BCE microphone is propagated through the user's head, which is unique in density, geometry, and bone-tissue ratio. It operates implicitly, while maintaining robustness across different behaviors. Extensive experiments involving 60 subjects demonstrate that HeadSonic achieves a commendable balanced accuracy of 96.59%, proving its efficacy and resilience against replay and synthesis attacks.
KW - Wearable authentication
KW - acoustic sensing
KW - biometrics
UR - https://www.scopus.com/pages/publications/105000196327
U2 - 10.1109/TMC.2025.3551272
DO - 10.1109/TMC.2025.3551272
M3 - Article
AN - SCOPUS:105000196327
SN - 1536-1233
VL - 24
SP - 7914
EP - 7928
JO - IEEE Transactions on Mobile Computing
JF - IEEE Transactions on Mobile Computing
IS - 9
ER -