A Response-Feedback-Based Strong PUF with Improved Strict Avalanche Criterion and Reliability.

Area of Science:

• Cybersecurity
• Computer Engineering
• Cryptography

Background:

• Physical Unclonable Functions (PUFs) are crucial for lightweight Internet of Things (IoT) authentication.
• Existing PUFs face vulnerabilities to machine learning (ML) modeling and statistical attacks.
• A balance between security and reliability in PUFs remains a challenge.

Purpose of the Study:

• To propose a robust response-feedback PUF resistant to ML attacks.
• To enhance the statistical properties (SAC and GSAC) of PUFs.
• To improve PUF reliability using an efficient method.

Main Methods:

• A novel PUF architecture combining Linear Feedback Shift Register (LFSR) and Arbiter PUF (APUF).
• Implementation of a Two-Level Reliability Improvement (TLRI) method.
• Evaluation of resistance against ML modeling attacks and assessment of statistical characteristics.

Main Results:

• The proposed PUF demonstrates resistance to existing ML modeling attacks.
• Achieved strong Strict Avalanche Criterion (SAC) and Generalized Strict Avalanche Criterion (GSAC) properties.
• The TLRI method provides 95% reliability with significantly reduced voting times and cycles.

Conclusions:

• The novel LFSR-APUF based PUF offers enhanced security and statistical properties for IoT authentication.
• The TLRI method significantly improves PUF reliability compared to traditional approaches.
• This work contributes to more secure and reliable IoT authentication solutions.