A stochastic approach to analog physical unclonable function
Troy Bryant, Sreeja Chowdhury, Domenic J. Forte, Mark Mohammad Tehranipoor, Nima Maghari · 2016
As cybersecurity becomes increasingly relevant with advancing technology, Physically Unclonable Functions (PUFs) have become a promising technique that utilize process variation and device mismatch to create unique device identifiers, secure keys, and other security features that are virtually impossible to duplicate. Many PUF circuits with diverse characteristics have been proposed, all of which exhibit a clear tradeoff between reliability, robustness, and size. This paper proposes a novel PUF which utilizes the nature of stochastic flash analog-to-digital (ADCs) to minimize hardware without sacrificing reliability or robustness against cloning. The proposed PUF creates a unique device identifier by employing the comparators used in a stochastic ADC, which are designed for maximum offset voltage due to device mismatch. The proposed PUF was modeled and simulated using a 130nm CMOS process. Simulation results show that the PUF has a normalized inter-hamming distance (uniqueness) of 48.5% and is reliable with a normalized average bit flip of 2-3% over ±10% variance in supply voltage and temperatures up to 100°C.