Leakage or Identification: Behavior-irrelevant User Identification Leveraging Leakage Current on Laptops

The convenience of laptops brings with it the risk of information leakage, and conventional security systems based on the password or the explicit biometric do little to alleviate this problem. Biometric identification based on anatomical features provides far stronger security; however, a lack of suitable sensors on laptops limits the applicability of this technology. In this paper, we developed a behavior-irrelevant user identification system applicable to laptops with a metal casing. The proposed scheme, referred to as LeakPrint, is based on leakage current, wherein the system uses an earphone to capture current leaking through the body and then transmits the corresponding signal to a server for identification. The user identification is achieved via denoising, dimension reduction, and feature extraction. Compared to other biometric identification methods, the proposed system is less dependent on external hardware and more robust to environmental noise. The experiments in real-world environments demonstrated that LeakPrint can verify user identity with high accuracy (93.6%), while providing effective defense against replay attacks (96.5%) and mimicry attacks (90.9%).