Multiple Observations for Secret-Key Binding with SRAM PUFs

Area of Science:

• Hardware Security
• Cryptography
• Integrated Circuit Design

Background:

• SRAM-PUFs (Static Random-Access Memory-Physical Unclonable Functions) are crucial for hardware security.
• Traditional helper data schemes for SRAM-PUFs rely on single observations, limiting performance.
• Modeling SRAM cell reliabilities is key to improving helper data schemes.

Purpose of the Study:

• Introduce a novel Multiple-Observations (MO) helper data scheme for SRAM-PUFs.
• Improve secret-key binding security and performance compared to existing methods.
• Investigate optimal strategies for helper data generation and reconstruction.

Main Methods:

• Developed a new MO helper data scheme using multiple SRAM-PUF enrollment observations.
• Proved the scheme's optimality in achieving secret-key capacity.
• Evaluated performance using Monte Carlo simulations with LDPC codes.
• Proposed and analyzed a new strategy for the Soft-Decision (SD) scheme using binary observations.
• Introduced a sequential update variation of the MO scheme.

Main Results:

• The MO scheme significantly improves performance by implicitly modeling SRAM cell reliabilities.
• The proposed SD scheme strategy achieves optimal performance using observable binary inputs.
• The sequential update MO scheme variation demonstrates improved error-correction over time.
• Both MO and the new SD strategy achieve optimal secret-key capacity.

Conclusions:

• The new MO helper data scheme offers superior performance and optimal secret-key capacity for SRAM-PUFs.
• The adapted SD scheme provides an effective alternative using readily available binary observations.
• Sequential updating of helper data further enhances the robustness of SRAM-PUF based key binding.