Application of probabilistic bits in measurement and sensing

Abstract

We introduce a data-driven measurement and sensing paradigm that capitalizes on the limited sensing capabilities of probabilistic bits (p-bits). Unlike traditional methods that rely on the high quality of individual devices, our approach achieves high precision through the extensive data collected from a large ensemble of p-bits. We demonstrate the feasibility of using magnetic tunnel junction-based p-bits in various applications, such as temperature stability monitoring, passive timekeeping, and weak magnetic field sensing. The paradigm is notably easy to implement and robust against device imperfections and non-uniformity, providing significant advantages in terms of practicality and scalability.