True Random Number Generation Using Latency Variations of FRAM

Rashid, Imtiaz; Ferdaus, Farah; Talukder, B. M. S. Bahar; Henny, Paul; Beal, Aubrey N.; Rahman, Mostafizur

doi:10.1109/tvlsi.2020.3018998

Cited by 22 publications

(14 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…TRNGs can be designed based on memory devices such as SRAMs [15][16][17], DRAMs [18], Flash RAMs [19], Ferro RAMs [20], STT MRAM [21,22], resistive RAMs and Memristors [23,24]. The opportunity is to leverage the potentially large and stochastic cellto-cell variations, considering that memory arrays contain extremely high numbers of addressable cells that can behave independently.…”

Section: Memory-based Trngsmentioning

confidence: 99%

“…To generate N random bits, l addresses of cells in the ReRAM array are needed, with l > N. Assuming that f bits are needed for each address, a stream of f x l bits is needed for the median scheme. For example, if the size of the array is 1,048,576 = 2 20 , then f = 20. A stream longer than N is needed for the protocol to handle the defective cells.…”

Section: Version 1: Design Of Trngs With the Median Schemementioning

confidence: 99%

See 1 more Smart Citation

TRNGs from Pre-Formed ReRAM Arrays

et al. 2021

View full text Add to dashboard Cite

Schemes generating cryptographic keys from arrays of pre-formed Resistive Random Access (ReRAM) cells, called memristors, can also be used for the design of fast true random number generators (TRNG’s) of exceptional quality, while consuming low levels of electric power. Natural randomness is formed in the large stochastic cell-to-cell variations in resistance values at low injected currents in the pre-formed range. The proposed TRNG scheme can be designed with three interconnected blocks: (i) a pseudo-random number generator that acts as an extended output function to generate a stream of addresses pointing randomly at the array of ReRAM cells; (ii) a method to read the resistance values of these cells with a low injected current, and to convert the values into a stream of random bits; and, if needed, (iii) a method to further enhance the randomness of this stream such as mathematical, Boolean, and cryptographic algorithms. The natural stochastic properties of the ReRAM cells in the pre-forming range, at low currents, have been analyzed and demonstrated by measuring a statistically significant number of cells. Various implementations of the TRNGs with ReRAM arrays are presented in this paper.

show abstract

Section: Memory-based Trngsmentioning

confidence: 99%

Section: Version 1: Design Of Trngs With the Median Schemementioning

confidence: 99%

TRNGs from Pre-Formed ReRAM Arrays

et al. 2021

View full text Add to dashboard Cite

show abstract

“…To filter out those temporally persistent cells using the proposed algorithm (described in Sect. III-C), we only chose those cells (set F C ) that are above the lower threshold range T h L = [15,23] (see Table I) for different chips used in the experiment. Although the number of eligible cells decreases after performing our proposed cell selection algorithm, the filtered cells are enough to generate high-quality random numbers.…”

Section: B Characterization Of Temporally Unbiased Cellsmentioning

confidence: 99%

“…There have been several high-quality and robust memorybased TRNGs, but most of them suffer from high-overhead and low-throughput [1], [3], [8], [20]- [22]. Therefore, several emerging memory-based TRNGs, capable of providing high density and throughput, have been proposed to overcome existing challenges [8], [12], [15], [22], [23]. Furthermore, MRAM-based TRNGs have gained attention because of their capability of generating significantly high quality and robust random numbers [12], [21], [24], [25].…”

Section: Introductionmentioning

confidence: 99%

True Random Number Generation using Latency Variations of Commercial MRAM Chips

Ferdaus¹,

Talukder²,

Sadi³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

The emerging magneto-resistive RAM (MRAM) has considerable potential to become a universal memory technology because of its several advantages: unlimited endurance, lower read/write latency, ultralow-power operation, high-density, and CMOS compatibility, etc. This paper will demonstrate an effective technique to generate random numbers from energy-efficient consumer-off-the-shelf (COTS) MRAM chips. In the proposed scheme, the inherent (intrinsic/extrinsic process variation) stochastic switching behavior of magnetic tunnel junctions (MTJs) is exploited by manipulating the write latency of COTS MRAM chips. This is the first system-level experimental implementation of true random number generator (TRNG) using COTS toggle MRAM technology to the best of our knowledge. The experimental results and subsequent NIST SP-800-22 suite test reveal that the proposed latency-based TRNG is acceptably fast (∼ 22M bit/s in the worst case) and robust over a wide range of operating conditions.

show abstract

“…Mulaosmanovic et al [28], [29] exploit the polarisation fluctuations due to domain wall motion, which is a Poisson process [30], in an HfO 2 based field-effect transistor. Other such TRNGs have recently been developed using ferroelectric random access memory (FRAM) technology [31]- [33]. However, none of these works exploit the randomness due to thermal fluctuations in the bistable well of a ferroelectric.…”

Section: Introductionmentioning

confidence: 99%

Kramers' escape problem for white noise driven switching in ferroelectrics

Ramesh¹,

Verma²,

Ajoy³

2021

Preprint

View full text Add to dashboard Cite

A simulation-based study of Kramers' escape problem in the bistable well of a ferroelectric capacitor is presented. This problem deals with the escape of a particle undergoing Brownian motion over an energy barrier. Using this framework under the assumption of homogeneous polarisation switching according to the Landau-Ginzburg-Devonshire (LGD) theory, two prospective device applications based on a ferroelectric capacitor are also outlined, namely true random number generation (TRNG) and stochastic resonance (SR). Our simulation results for the former demonstrate that by adding white noise and an appropriately tuned voltage pulse to the material system, it is possible to facilitate controlled, probabilistic switching between its two stable polarisation states. We predict that this could provide the theoretical framework for practical implementations of TRNGs. In addition, we also develop a theory for stochastic resonance in a ferroelectric using linear response theory. We show, analytically and via numerical simulations that the addition of an optimal amount of noise to a weak periodic signal given as an input to a non-linear system can enhance its detection.

show abstract

True Random Number Generation Using Latency Variations of FRAM

Cited by 22 publications

References 32 publications

TRNGs from Pre-Formed ReRAM Arrays

TRNGs from Pre-Formed ReRAM Arrays

True Random Number Generation using Latency Variations of Commercial MRAM Chips

Kramers' escape problem for white noise driven switching in ferroelectrics

Contact Info

Product

Resources

About