A 0.7 pJ/bit, 1.5 Gbps Energy-Efficient Image-Based True Random Number Generator. Issue 3 (3rd April 2023)

• Record Type: Journal Article
• Authors: Kumar, Dhirendra; Mankali, Lakshmi Likhitha; Misra, Prasanna Kumar; Goswami, Manish
• Is Part Of: IETE journal of research. Volume 69:Issue 3(2023)
• Page Start: 1260
• Page End: 1270
• Publication Date: 2023-04-03
• Subjects: Entropy -- LFSR -- Meta-stable -- NIST -- Random number -- TRNG; Electronics -- Periodicals; Telecommunication -- Periodicals; Electronics; Telecommunication; Periodicals; 621.38
• DOI: 10.1080/03772063.2020.1859957
• Languages: English
• ISSNs: 0377-2063

Abstract

ABSTRACT: Random numbers cover a vast spectrum of applications. Hence generating it effectively with high performance is the need of the hour. This paper presents the novel design and implementation of high performance and energy-efficient true random number generator (TRNG) using images as a source. In the proposed work, the harvesting mechanism that comprises of hashing technique block (HTB) is used to reduce the intensity of pattern, an event counting circuit (ECC) is used for comparing the events, while a linear feedback shift register (LFSR), designed by considering a primitive polynomial function, is used to obtain the random numbers. The addition of 8 × 1 multiplexer (MUX) with meta-stable circuit feeder control lines had further increased the unpredictability in the proposed system. The implementation of the work has been done in the Xilinx Vivado simulation tool followed by the Cadence Virtuoso circuit simulation environment. The maximum speed of 1.5 Gbps with power dissipation of 1 mW and 0.7 pJ/bit energy efficiency with a layout area of 2218 μm 2 has been achieved in this work. NIST 800.22 statistical test suite and uniformity test comprising of Kolmogorov–Smirnov and Chi-square test have also been performed for validation of generated random numbers. The obtained binary sequences have passed all tests successfully with calculated entropy up to 0.999999999. The autocorrelation factor (ACF) of the output bit streams has been obtained as approximately zero (∼0) within 96% confidence bounds of a Gaussian distribution ( µ = 0, 3σ). The proposed design is thus suitable for true random number generation.