University of Michigan
Electrical & Computer Engineering
1301 Beal Ave., 2417C EECS
Ann Arbor, MI 48109
Tel: 734 763 4526
Fax: 734 763-4617
STMicroelectronics is gratefully acknowledged for IC fabrication.
Random number generation plays a crucial role in cryptography and security. For example, public key cryptography systems demand strong key pair generation to ensure a third-party cannot decrypt secret messages. Traditionally random bit sequences are generated in digital systems using pseudo-random number generators, which produce sequences that are not truly random but contain exploitable patterns, such as repetition and correlation. True random number generators (tRNGs) use physical phenomena as a randomness source.
A true random number generator (tRNG) is proposed that, for the first time, uses the random physical process of time to oxide breakdown under voltage stress. Time to breakdown is repeatedly measured with a counter and serialized into a bitstream. The 1200 um^2 tRNG, called OxiGen, was fabricated in 65 nm CMOS, passes all 15 NIST randomness tests without post-processing and in a 3 month run generated sufficient bits for worst-case expected internet use while being < 10% exhausted.
Nurrachman Liu, Nathaniel Pinckney, Scott Hanson, Dennis Sylvester, David Blaauw, “A True Random Number Generator using Time-Dependent Dielectric Breakdown,” IEEE Symposium on VLSI Circuits (VLSI-Symp), June 2011 ©IEEE