STT-DPSA: Digital PUF-Based Secure AuthenticationUsing STT-MRAM for the Internet of Things

Area of Science:

• Hardware Security
• Integrated Circuits
• Emerging Memory Technologies

Background:

• Physical Unclonable Functions (PUFs) leverage manufacturing variability for unique IC identification.
• Analog PUFs lack robustness against environmental variations, posing security risks.
• Developing reliable and secure authentication for Internet of Things (IoT) devices is critical.

Purpose of the Study:

• To propose a novel digital PUF-based secure authentication model, STT-DPSA, utilizing Spin-Transfer Torque Magnetic Random-Access Memory (STT-MRAM).
• To design a computationally lightweight and environmentally robust authentication architecture for IoT devices.
• To enhance hardware security by addressing the limitations of existing PUF technologies.

Main Methods:

• Developed a digital PUF model (STT-DPSA) using STT-MRAM, implementing matrix multiplication or stochastic logic operations.
• Validated PUF reliability across a temperature range of -35 °C to 110 °C and Vdd range of 0.95 V to 1.16 V.
• Implemented STT-DPSA using FPGA design flow with specific parameters (n=32, i=o=1024, k=8, l=2) and evaluated using Synopsys Design Compiler with TSMC 0.18 um process.

Main Results:

• STT-DPSA demonstrates robustness against environmental conditions, functioning reliably within specified temperature and voltage intervals.
• The proposed model achieves a high level of security, requiring an attacker an estimated time complexity of O(2^256) to compromise.
• The architecture is computationally lightweight, suitable for resource-constrained IoT devices.

Conclusions:

• The STT-DPSA model offers a viable and secure authentication solution for IoT devices, overcoming the environmental vulnerabilities of analog PUFs.
• The use of STT-MRAM in digital PUFs provides a promising direction for robust and efficient hardware security.
• The proposed architecture balances hardware efficiency, security, and environmental resilience.