A Novel FeFET Array-Based PUF: Co-optimization of Entropy Source and CRP Generation for Enhanced Robustness in IoT Security
Hanyong Shao, Yuejia Zhou, Weiqin Huang, Chang Su, Zhiyuan Fu, Wenpu Luo, Kechao Tang, Ru Huang
Abstract
Enhancing the resistance of Physical Unclonable Function (PUF) to machine learning (ML) attacks while ensuring its reliability remains a critical challenge. In this work, we propose a novel strong PUF based on a 10×10 FeFET array, delivering high security, reliability, and reconfigurability. For the first time, the entropy source and CRP generation of PUF are co-optimized for robustness. We utilize the high variation in ferroelectric (FE) minor loop and high uniformity in major loop as reconfigurable static entropy source, then adopt non-linear Hamming distance comparison and window comparator for the generation of CPR. As a result, our approach reduces the accuracy of ML attacks by 20% compared to RRAM PUF, while keeping the raw bit error rate (BER) down to 1.7% at 100°C. Due to the high endurance of FeFETs, the capacity of reconfigurability reaches 1 × 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">8</sup> . The PUF fits well for security applications due to its robustness and energy-efficiency.