On the use of an RTN simulator to explore the quality trade-offs of a novel RTN-based PUF

Abstract

Physical Unclonable Functions (PUFs) use variability as an entropy source from which to generate secure authentication and identification. While most silicon PUFs exploit the well-known Time-Zero Variability of CMOS technologies, the lack of efficient simulation tools for the Time-Dependent Variability (TDV) has left the potential benefits of this other kind of variability largely unexplored. However, recent advances in the field are allowing this exploration to begin. The objective of this paper is then to take a recently reported simulation tool to design a novel PUF that uses the Random Telegraph Noise (RTN), a TDV phenomenon, as the underlying entropy source. In the ensuing analysis, essential design guidelines are provided to best exploit such entropy source with factors like transistor biasing and sizing.

Design considerations for a CMOS 65-nm RTN-based PUF

“…A relationship between transistor size and the optimum value of the current ICC is defined. Since the transistors used in this work have the size ratio W/L = 80nm/60nm, following the indications described in [10], the optimum ICC should be 100nA. All in all, the MVF captures, as the MCF does, the RTN in the transistor by subtracting the maximum envelope (CMAXV) and the minimum envelope (CMINV) over a period of time.…”

“…The quality of the RTN-PUF has been demonstrated in [6] and design preliminaries were discussed in [10] [11]. It was found that, for the 65-nm CMOS technology used as a case study, minimum-size PMOS transistors (W/L = 80nm/60nm) yielded a better PUF response quality.…”