High-speed true random number generation based on paired memristors for security electronics

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• True random number generators (TRNGs) are vital for hardware security, especially in mobile computing and IoT devices.
• Intrinsic variations in memristors, typically undesirable, are leveraged as a novel entropy source.
• Existing TRNGs face challenges in efficiency and security for modern applications.

Purpose of the Study:

• To demonstrate a novel TRNG based on the inherent resistance variations of memristors.
• To explore the potential of tantalum oxide-based memristors for generating physically-implemented random numbers.
• To improve the unbiasedness and throughput of random number generation using memristor technology.

Main Methods:

• Utilized tantalum oxide-based memristors (Pt/TaOx/Ta) as the entropy source.
• Employed cyclic switching and comparison of memristor off-state resistance values.
• Implemented an alternating read scheme within the TRNG circuit design.

Main Results:

• Achieved random bit generation by exploiting pronounced off-state resistance variations.
• Significantly improved the unbiasedness of generated random numbers through the alternating read scheme.
• Demonstrated that the generated bitstream successfully passed standard randomness tests.
• Fabricated memristors exhibited fast programming/erasing speeds of approximately 30 ns.

Conclusions:

• Intrinsic memristor variation offers a viable and efficient entropy source for TRNGs.
• The proposed TRNG design shows potential for high-throughput, physically-implemented random number generation.
• This approach contributes a promising solution for enhancing hardware security in electronic devices.