Tractable, Not Truncated: Colored-Noise TRNG Security via Fractional Brownian Motion

What it is

A preprint (IACR ePrint 2026/2662, approved 28 October 2026) by Maciej Skórski, Nathalie Bochard and Licinius Benea. It gives a security model for oscillator-based true random number generators (TRNGs) that treats thermal (white) and flicker noise together. The authors say flicker noise matters more as transistors shrink.

How it works

The oscillator phase is modeled as fractional Brownian motion. The authors prove a quasi-renewal property: the phase uncertainty left after conditioning on the entire past depends only on the elapsed time. That gives closed-form worst-case min-entropy and sampling requirements. The authors say existing models discard or truncate flicker's long memory, which overstates security.

For parameter estimation, the paper extends spectral and Allan-variance analysis to finite samples. The abstract says the noise parameters are estimated consistently from one timing trace, and that the path from measurement to sampling requirement is closed-form and needs no simulation.

What to do

The abstract gives no procedure, code or constants, so the actionable points are conclusions:

• In the FPGA and ASIC experiments, flicker variance meets white-noise variance within about 100 cycles.
• Flicker accounts for more than 99% of the uncertainty behind a 0.99 min-entropy target.
• Designs that rely only on thermal noise need adjustment or redesign. Counting flicker noise allows a bitrate two to three orders of magnitude higher.

Caveats

Only the abstract was available. The specific platforms, the estimator formulas and the assumptions behind the quasi-renewal proof are not stated. The page lists the paper as a preprint with no peer-review venue.