Skip to content
Standards mapping

Every layer has a standard behind it.

An entropy certificate covers a component. A complete random bit generator is a construction. Here is where each standard applies, and where Quside’s evidence lands in it.

The relationship

Source, construction, mechanism.

Read from the bottom up. The entropy source produces unpredictability, the construction describes how it seeds a deterministic mechanism, and the mechanism produces what the application actually calls. Each layer has its own standard, and each has its own boundary.

Application

Consumes random values

Key generation, nonces, session material, tokens. The application almost never talks to a physical source directly.

Deterministic mechanism

Expands seed material

An approved DRBG turns seed material into the stream the application consumes. A sound construction whose security depends on strong seeding, correct construction and appropriate reseeding.

SP 800-90A

Construction

Connects the two halves

Defines how a validated entropy source and an approved deterministic mechanism are assembled into a complete random bit generator, including how and when the mechanism is seeded.

SP 800-90C

Entropy source

Produces the unpredictability

The physical process, its documented model, the entropy quantified within that model, and the continuous health tests the standard requires. This is where certificate E178 sits, scoped to Quside PCIe One and FMC One.

SP 800-90B
The landscape

What each document is for.

The groups that matter here: the NIST SP 800-90 series that governs random bit generation, the post-quantum algorithm standards finalised in 2024, and the European evaluation schemes.

NIST SP 800-90 series

SP 800-90A

Deterministic mechanisms

Specifies approved deterministic random bit generator mechanisms — the layer your application usually calls. Its security depends on being seeded and reseeded from a trusted entropy source.

SP 800-90B

Entropy sources

Sets requirements for entropy sources and the methodology used to validate them, including the documented noise-source model, entropy estimation and continuous health tests. Quside holds entropy certificate E178 under this document, scoped to Quside PCIe One and FMC One.

SP 800-90C

Constructions

Final in September 2025. Describes how a validated entropy source and an approved deterministic mechanism combine into a complete random bit generator. Before it, the two halves were specified separately and assembled by convention.

Post-quantum algorithms

FIPS 203

ML-KEM

Key encapsulation, finalised August 2024. Key generation and encapsulation both consume randomness.

FIPS 204

Signature scheme

One of the two signature standards finalised in August 2024.

FIPS 205

Signature scheme

The second signature standard finalised in August 2024.

European schemes

ENS

Spanish national security framework

Categorises the systems the Spanish public sector runs as Básica, Media or Alta, and qualifies products for use up to a given category. Quside PCIe One and FMC One are qualified at Alta, the highest of the three, which covers the two below it.

BSI AIS 20/31

German evaluation methodology

Defines functionality classes for deterministic and physical random number generators, expecting a stochastic model of the noise source and online testing for physical generators. Materials were updated across 2024–2025.

Common Criteria

ISO/IEC 15408

Evaluates a defined target of evaluation to a stated assurance depth. It is a framework for how thoroughly a product is examined, not a grade for randomness in the abstract; the RNG requirements inside a European evaluation typically come from AIS 20/31.

Post-quantum algorithms

A different threat, a different layer

FIPS 203, 204 and 205, finalised in August 2024, address the risk that a sufficiently capable quantum computer could break the public-key cryptography in use today. They change which algorithms a system runs.

Every one of those schemes consumes randomness — key generation, encapsulation, signature nonces — and the algorithm layer has no way to report a source that has quietly drifted.

Entropy assurance

An older problem, still open

Entropy assurance addresses whether the unpredictability those algorithms consume is actually unpredictable, and whether it stays that way once a device is in the field.

PQC algorithms and entropy assurance solve different parts of the same cryptographic architecture. Neither substitutes for the other, and quantum entropy does not make a system post-quantum secure — it is quantum entropy that strengthens classical and post-quantum cryptographic systems alike.

Cryptographic renewal is a rare moment when the physical foundation is on the table rather than fixed. Products are being redesigned anyway. Three properties worth writing into the specification: that the entropy source is local to the device, that its health is verified at runtime rather than asserted at procurement, and that its validation scope is stated precisely enough to know what it covers.

Direction of travel

What regulators have actually said.

Procurement direction is frequently repeated as though it were law. These are the signals, with what each one does and does not commit you to.

United States

FIPS certification paths increasingly depend on SP 800-90B entropy evidence rather than treating the entropy source as an implementation detail.

Caveat

This raises the evidentiary bar for the source. It does not make a validated source sufficient on its own — SP 800-90C describes what a complete random bit generator requires.

France

ANSSI has stated that post-quantum cryptography should enter product qualification from 2027, and that buying products without PQC after 2030 will not be reasonable.

Caveat

This is a strong signal about procurement direction. It is not a universal legal ban, and it should never be presented as one.

Germany

BSI AIS 20/31 materials were updated across 2024–2025, keeping a stochastic model of the noise source and online testing at the centre of physical generator evaluation.

Caveat

AIS 20/31 sets requirements for the generator itself. It does not define how source health is surfaced to an operator once the device is in the field.

Mapping a requirement to a design?

Send the clause you have to satisfy. We will say which layer it lands on, what our evidence covers at that layer, and what you or an evaluator would still produce.