| Autor: |
Kouta Ibukuro, Fayong Liu, Muhammad Khaled Husain, Moïse Sotto, Joseph Hillier, Zuo Li, Isao Tomita, Yoshishige Tsuchiya, Harvey Rutt, Shinichi Saito |
| Jazyk: |
angličtina |
| Rok vydání: |
2020 |
| Předmět: |
|
| Zdroj: |
AIP Advances, Vol 10, Iss 11, Pp 115101-115101-12 (2020) |
| Druh dokumentu: |
article |
| ISSN: |
2158-3226 |
| DOI: |
10.1063/5.0023647 |
| Popis: |
The need for hardware random number generators (HRNGs) that can be integrated in a silicon (Si) complementary-metal–oxide–semiconductor (CMOS) platform has become increasingly important in the era of the Internet-of-Things (IoT). Si MOSFETs exhibiting random telegraph signals (RTSs) have been considered as such a candidate for HRNG, though its application has been hindered by RTS’s variability and uncontrollable, unpredictable characteristics. In this paper, we report the generation and randomness evaluation of random numbers from RTSs in a Si single electron pump (SEP) device at room temperature. SEP devices are known to consistently produce RTSs due to a quantum dot electrically defined by multi-layer polycrystalline Si gates. Using RTSs observed in our devices, random numbers were extracted by a classifier supported by supervised learning, where part of data was used to train the classifier before it is applied to the rest to generate random numbers. The random numbers generated from RTSs were used as inputs for the Monte Carlo method to calculate the values of π, and the distribution was compared against the result obtained from the Mersenne Twister, a representative pseudo-random number generator (PRNG), under the same condition. π was estimated more than 80 000 times, and the distribution of the estimated values has a central value of 3.14 with a variance of 0.273, which is only twice as large as the result from PRNG. Our result paves a way to fully electronic CMOS compatible HRNGs that can be integrated in a modern system-on-a-chip in IoT devices. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
