Abstract

Analog Physically Unclonable Functions is a new class of Physically Unclonable Functions with the characteristic that the output is not a bitword (as it is in the traditional Physically Unclonable Functions), but an analog value that is successively converted to digital by a suitable circuit called digitizer. The advantage is that, potentially, if a suitable digitizer is employed, a single cell can produce more than one bit, simplfying the implementation and reducing the footprint. This paper proposes a simple digitizer for Gaussian Analog Physically Unclonable Functions that exploits the properties of quantization noise to obtain a stable and uniformly distributed output.