MetaSPICE: Metaprogramming SPICE Framework for the Design Space Exploration of PUF Circuits
Athanasios Xynos, Vasileios Tenentes · 2023
Silicon Physical Unclonable Functions (PUFs) generate challenge-response pairs that are unique to each die, allowing for secure identification, authentication and cryptographic keys generation. However, designing and evaluating PUFs is a complex task that can require a vast amount of SPICE netlists simulation. In this paper, we present MetaSPICE, which is a novel metaprogramming SPICE framework for the design space exploration of PUFs. Metaprogramming refers to the process of writing programs that manipulate other programs, data structures or even themselves. The designer in MetaSPICE, instead of defining multiple SPICE netlists, has to define a single SPICE template that is used to generate multiple netlists. The MetaSPICE is complementary to classical SPICE simulators, and enables the design exploration impact on PUFs’ performance. The framework consists of a template engine for generating SPICE models from SPICE templates, and software for evaluating the performance of PUF circuits in terms of uniqueness, uniformity and reliability from SPICE simulation results. We demonstrate the effectiveness of MetaSPICE by evaluating the performance and presenting improvements on a recently proposed PUF design.