Biometrics for Electronic Eyes: System Authentication with Embedded CMOS Image Sensor

Abstract

Identically designed electronic devices such as transistors may have random variations in their physical quantities. Inspired by human biometrics, this paper presents a novel approach that enables home appliances to have unique identities based on the intrinsic features of a CMOS image sensor (CIS) embedded in their electronic eyes (i.e., in-built cameras). More specifically, the column component of the fixed pattern noise (FPN) that inevitably exists in most modern CISs is utilized as a unique feature of the CIS for describing a unique ID for the embedded system. To practically capture column FPNs that were originally canceled out by a correlated double sampling (CDS) process, this paper also proposes a novel readout scheme. The feasibility of the proposed CIS-based identification is assessed with real measurements using 12 CIS prototypes. In conclusion, a set of applications requiring unique device identifications are presented for demonstrating the utility of the proposed approach in the domain of consumer electronics.(1)