Impact of temperature fluctuations on SRAM- based physically unclonable functions

Abstract

Hardware-based security systems are gaining attention owing to advancements in Internet of things (IoT) and mobile technologies. Physically unclonable functions (PUFs) are hardware security primitives that act as semiconductor device fingerprints and can be applied to chip authentication, anti-counterfeiting, and secure data storage. Designers of security architectures have especially been focusing on static random-access memory (SRAM)-based PUFs because of their widespread applicability in systems on a chip (SoC) and embedded systems. However, a primary concern in the use of SRAM-based PUFs is their low robustness. Although earlier research studies have already determined that SRAM-based PUFs exhibit bit error rates (BERs) of approximately 5% under ideal conditions, it remains essential to evaluate robustness under varied environmental factors such as temperature and voltage conditions. In this work, the impact of temperature variations on BER is investigated. It is observed that temperature fluctuations increase BER, potentially leading to chip misidentification. The increased BER in an SRAM unit cell under temperature fluctuations is attributed to variations in the threshold voltage, referred to as the threshold voltage temperature coefficient (TVTC) phenomenon. The TVTC phenomenon is unavoidable because it is generated by inevitable process variations. Monte Carlo simulations on process variations reveal that the TVTC phenomenon increases BER to 3% points as temperature is increased from room temperature to 80 degrees C. Therefore, it should be noted that the BERs of SRAM-based PUFs are increased when temperature fluctuations are applied.