Physical Random Number Generators

Michael Gude

Werner Schindler

Tatsuro Sugiura, Yuki Yamanashi, Nobuyuki Yoshikawa

A physical random number generator, which generates truly random number trains by using the randomness of physical phenomena, is widely used in the field of cryptographic applications. We have developed an ultra high-speed superconductive physical random number generator that can generate random numbers at a frequency of more than 10GHz by utilizing the high-speed operation and high-sensitivity of superconductive integrated circuits. In this study, we have statistically evaluated the quality of the random number trains generated by the superconductive physical random number generator. The performances of the statistical tests were based on a test method provided by National Institute of Standards and Technology (NIST). These statistical tests comprised several fundamental tests that were performed to evaluate the random number trains for their utilization in practical cryptographic applications. We have generated 230 random number trains consisting of 20, 000-bits by using the superconductive physical random number generator fabricated by the SRL 2.5kA/cm2 Nb standard process. The generated random number trains passed all the fundamental statistical tests. This result indicates that the superconductive random number generator can be sufficiently utilized in practical applications.

Mario Stipčević, Çetin Kaya Koç

Yuki Yamanashi, Nobuyuki Yoshikawa

A novel high-speed physical random number generator using the superconductive single-flux-quantum (SFQ) circuits and thermal noises in the circuit has been proposed. The proposed physical random number generator is similar to an SFQ balanced comparator. Thermal noises in shunt resisters are used to obtain random outputs. Because of the high-sensitivity of SFQ circuits, the true random numbers can be generated without amplification of the noises at high clock frequency. Generation of the true random numbers at the frequency of several tens GHz can be realized by using the proposed circuit. Moreover, the circuit scale of the proposed circuit is very small and the generation frequency of the random number generator can be easily enhanced by parallelizing the circuits. We have designed and tested the superconductive physical random number generator using the 2.5 kA/cm2Nb standard process and evaluated the quality of the generated random number train. We have examined a 220-bit random number train generated at a low frequency and no periodicity is observed. We believe that the superconductive physical random number generator can be applied to cryptographic applications for more secure information processing.

Zhiliang L. Yuan