Optically reconfigurable physical unclonable functions based on 2D MoS₂ ring-oscillator arrays for attack-resistant...

Area of Science:

• Materials Science and Engineering
• Cybersecurity
• Nanotechnology

Background:

• Hardware-level security is critical for the Internet-of-Things (IoT) and edge computing.
• Physical Unclonable Functions (PUFs) offer unique identifiers but often lack reconfigurability and are vulnerable to machine learning attacks.
• Existing silicon-based PUFs have limitations in entropy and dynamic security capabilities.

Purpose of the Study:

• To present an optically reconfigurable physical unclonable function (PUF) for enhanced hardware security.
• To leverage molybdenum disulfide (MoS2) and optical stimuli for dynamic rekeying capabilities.
• To develop and validate a robust key-generation pipeline resilient to machine learning attacks.

Main Methods:

• Fabrication of a 64-cell array of five-stage ring oscillators using wafer-scale monolayer MoS2.
• Exploitation of spectrally selective frequency shifts under Red, Green, and Blue (RGB) illumination for optical reconfigurability.
• Implementation of a key-generation pipeline including normalization, q-ary quantization, SHA-256 privacy amplification, and HMAC-based key derivation.

Main Results:

• Demonstrated spectrally selective frequency shifts enabling dynamic optical entropy and reversible rekeying.
• Achieved near-ideal key uniformity (~50%) and high inter-device Hamming distances.
• Showcased resilience against machine learning attacks (≤ 52% accuracy) and successful image encryption/authentication.

Conclusions:

• Introduced a novel class of material-intrinsic, optically addressable security primitives.
• The optically reconfigurable MoS2 PUF provides a promising solution for trusted edge computing.
• This approach enhances security and reconfigurability in IoT devices without hardware modification.