A Design of Reliable True Random Number Generator for Cryptographic Applications

Abstract

The scheme of a device that should have a simple and reliable implementation and that, under simply verifiable conditions, should generate a true random binary sequence is defined. Some tricks are used to suppress bias and correlation so that the desired statistical properties are obtained without using any pseudorandom transformation. The proposed scheme is well represented by an analytic model that describes the system behaviour both under normal conditions and when different failures occur. Within the model, it is shown that the system is robust to changes in the circuit parameters. Furthermore, a test procedure can be defined to verify the correct operation of the generator without performing any statistical analysis of its output.

Design of Testable Random Bit Generators

“…Nevertheless, the additional hypothesis of independence of each raw bit allows to adopt the counting of the transitions in the sequence s[i] as a straightforward entropy estimator (Figure 2) for the source [6]:…”

A Design for a Physical RNG with Robust Entropy Estimators

“…The amplified noise is also modelled as a stationary stochastic process, and the autocorrelation function of the random numbers are computed. We mention that unlike the present paper reference [4] yet considers idealized assumptions (Gaussian white noise etc. ), which clearly simplify analysis.…”

“…In [4] a design of a physical RNG is investigated that also exploits the switchings of a comparator. The amplified noise is also modelled as a stationary stochastic process, and the autocorrelation function of the random numbers are computed.…”

“…), which clearly simplify analysis. [4] exploits the number of comparator switchings within fixed time periods for an online test (cf. Sect.…”

“…In the last years several designs of physical RNGs have been proposed [4,5,6,7,9] etc., and several evaluation guidances and standards were developed and became effective [1,2,11,13,17]. These documents define properties that strong RNGs should fulfil, and the evaluation guidances explain how these criteria shall be verfied.…”