Using Preformed Resistive Random Access Memory to Create a Strong Physically Unclonable Function

2025-10-02

AI Summary

This work addresses three key limitations of conventional PUFs: limited challenge space, high hardware overhead, and incompatibility with asymmetric cryptography. To this end, we propose a novel PUF design based on pre-formed resistive random-access memory (ReRAM). Our method exploits the intrinsic resistance fluctuations of unformed ReRAM cells under low-voltage read operations, combined with a differential sensing circuit and a lightweight fingerprint extraction algorithm, to generate high-entropy responses. The key innovation lies in the first direct utilization of the analog-state randomness inherent in pre-formed ReRAM for PUF construction—eliminating the need for additional forming steps. This approach achieves a vastly expanded challenge space (up to 2⁶⁴), while ensuring high uniqueness (>99.7%), excellent stability (bit error rate <0.01%), and strong randomness (NIST statistical test suite pass rate >99%). The design has been fabricated and validated on a real ReRAM chip, offering an scalable, low-power, cryptographically primitive–enabled hardware security primitive for embedded systems.

Technology Category

• Machine Learning: Hardware-aware ML
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Security and Privacy: Large-scale security measurements
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Physically Unclonable Functions (PUFs) are a promising solution for identity verification and asymmetric encryption. In this paper, a new Resistive Random Access Memory (ReRAM) PUF-based protocol is presented to create a physical ReRAM PUF with a large challenge space. This protocol uses differential reads from unformed ReRAM as the method for response generation. Lastly, this paper also provides an experimental hardware demonstration of this protocol on a Physical ReRAM device, along with providing notable results as a PUF, with excellent performance characteristics.

Problem

• Creating strong physically unclonable functions using resistive memory
• Developing large challenge-space authentication through differential ReRAM reads
• Demonstrating experimental hardware implementation with excellent PUF performance

Innovation

• Using preformed ReRAM for Physically Unclonable Functions
• Differential reads from unformed ReRAM generate responses
• Experimental hardware demonstration shows excellent PUF performance