Memristive True Random Number Generator with Intrinsic Two-Dimensional Physical Unclonable Function

The development of physical-level primitives for cryptographic applications has emerged as a trend in the electronic community, while the methods for protecting the generators from counterfeiting have yet to be explored. In this study, two-dimensional electronic fingerprinting was demonstrated and integrated into a memristive true random number generator (TRNG). For the device function of the TRNG, two modes of primitives are presented, and the physical entropy sources are analyzed via a recurrent neural network, which is resilient for machine learning prediction. For anticounterfeiting of the device, a two-dimensional physical unclonable function (PUF) could provide a high entropy value and multiple verification codes. Because of its extremely high surface-to-volume ratio, high sensitivity to the environment, inevitable randomness introduced in the fabrication process, and the ability to be transferred onto arbitrary substrates (easy to integrate into a single device), this two-dimensional PUF device could be a general solution for anticounterfeiting of nanoelectronics.