Threshold voltage based Memristor Crossbar PUF

Abstract

While technologies are adopting smart objects, which are moving toward the Internet of Things (IoT), data security and privacy are becoming essential. IoT security plays a crucial role in a world where smart devices are interconnected at every level, from wearable to home and building automation to smart cities and infrastructure, to smart industries, and to smart things. Since the interconnection involves all types of IoT devices, any vulnerability in one of these devices allows intruders to gain access to the network. On the other hand, many of those devices lack security layers due to low marginal cost constraints. Also, some of these devices are power-constrained and operate on harvested energy, hence requiring low design complexity. This opens doors for adversaries to exploit vulnerabilities in the most primitive devices and create access channels to the network. So this makes security a big challenge to hardware developers. A low-power, low-cost, and low-complexity security overhead solution must be integrated to ensure adequate security resilience. Physical Unclonable Function (PUF) comes in handy as an emerging lightweight hardware security primitive for key management and device authentication. PUFs utilize the permenant physical fingerprint of manufactured devices, originating from stochastic variations that are randomly generated during the manufacturing process of those devices, which cannot be reproduced even if the process is the same as that of the original manufacturer. This variation in the manufacturing process gives the produced devices unique physical properties. PUF exploits those unique, uncontrolled device physical parameters. The set of input and output vectors in PUF is called challenge-response pairs (CRP). The mapping of the CRPs of different PUF (devices) Instances are expected to be different from each other. Many silicon-based PUFs have been proposed over the years. One of the emerging PUF designs is based on a memristor due to the advantages of low power, area, and security. A memristor is a device that was discovered by Chua. The device stores digital data in the form of two states: high resistance ($R_{OFF}$) and low resistance ($R_{ON}$). Each of the states can be reached when an applied electric voltage across its terminals crosses their corresponding threshold voltages. Recently, memristors crossbars were used as PUF by exploiting $R_{ON}$ and $R_{OFF}$ errors to generate a random output bit stream. These crossbars use comparators to select PUF output based on the voltage difference. Comparators are known to be power-hungry components and occupy a relatively large area. We propose a comparator-less crossbar PUF that is based on the memristor's threshold voltage entropy, which is due to manufacturing process variations. Sneak path current is the main challenge of crossbars. It adds a parallel equivalent resistance to the selected one, which causes wrong readings. We propose uni-directional memristors and demonstrate that our proposed crossbar PUF is a sneak path resistance. A transparent PUF core output can be modeled for attacks. An extra security layer, which is low-power, is integrated into the system to scramble PUF output.

Date of Award	2023
Original language	American English
Awarding Institution	HBKU College of Science and Engineering