A Key Storage Scheme Based on Noisy SRAM PUF Fingerprints and Fuzzy Vault

Abstract

Securing secret key storage in resource-constrained embedded integrated circuits (ICs) is a core challenge in hardware security and IC testing. Physical Unclonable Functions (PUFs) offer a lightweight hardware primitive for this task, yet traditional schemes either only utilize stable PUF responses or discard unstable noisy cells as useless data. This paper proposes a novel Static Random Access Memory (SRAM)-PUF-based key storage scheme that integrates Fuzzy Vault (FV) construction with both Stable Cell Fingerprints (SCF) and Unstable Cell Fingerprints (UCF) of SRAM chips. Distinguished from extractor-centric conventional approaches, the proposed scheme embeds externally provisioned or dynamically updated cryptographic keys into the FV, enabling flexible and scalable key management for embedded ICs. Moreover, a noise-cell-based template mechanism is designed to exploit the statistical skew characteristic of uns table SRAM cells, converting previously discarded noise into an additional entropy source to enhance key storage security. Experimental validation on two commercial SRAM chip series (Hynix HY62256A and ST WS62256LLP-70) demonstrates that after device screening and parameter optimization, the scheme achieves a 100% Key Reconstruction Rate (KRR) under laboratory conditions. The results verify the feasibility of fusing SCF and UCF in a unified FV framework, providing an error-tolerant, updateable, and highly reliable solution for secure key storage in embedded ICs with significant practical value for hardware security testing and validation of IC products.