The vocabulary, used precisely.
Entropy, random bit generation, certification and post-quantum terms — defined briefly, with the boundaries stated. Where a term deserves a longer answer, the entry links to the page that gives it one.
A
Top- AIS 20/31BSI evaluation methodology for random number generators
The German federal office for information security (BSI) methodology used to evaluate random number generators inside Common Criteria evaluations. It classifies physical, hybrid and deterministic generators, and its supporting materials were updated across 2024 and 2025.
C
Top- Chiplet
A die that implements part of a system and is packaged together with other dies rather than integrated onto a single monolithic chip. It is one of the ways an entropy source can move from a module into a customer’s own package.
- Common Criteria
An international framework for evaluating security products against a defined security target, carried out by accredited laboratories. Random number generation inside a Common Criteria evaluation in Germany is assessed using BSI AIS 20/31.
- Conditioning
Post-processing applied to the raw output of an entropy source so that the entropy present is distributed more evenly across the output bits. Conditioning redistributes entropy; it does not create any.
- Crypto agility
The ability to change cryptographic algorithms, parameters or key sizes in a deployed product without redesigning it. It matters most in long-lived equipment that will outlive the cryptography it shipped with.
D
Top- DRBGDeterministic random bit generator
An algorithm that expands a seed into a much longer output stream, specified in NIST SP 800-90A. Approved DRBGs depend on strong seeding, correct construction and appropriate reseeding from trusted entropy sources — which is the dependency an entropy source exists to satisfy.
E
Top- E178
The NIST SP 800-90B entropy certificate held by Quside. Its scope is Quside PCIe One and FMC One. It validates the entropy source — not a complete random bit generator, and not the systems built on top of it.
- ENSEsquema Nacional de Seguridad
The Spanish national security framework. Systems are categorised básica, media or alta according to the impact of a breach, and products are qualified for use up to a given category. Quside PCIe One and FMC One are qualified at Alta, the highest of the three, which therefore also covers the two below it. It is a qualification rather than a certification, and it is scoped to version 1.0 of those two products.
- Entropy
The unpredictability a cryptographic system consumes. Every key, nonce, seed and token depends on it, and it has to originate in a physical process somewhere in the system.
- Entropy source
In SP 800-90B terms, the complete assembly that produces raw randomness: the noise source, its digitisation, the health tests that watch it, and any conditioning applied afterwards. It is a component of a random bit generator, not a finished one.
F
Top- FIPS 203
The NIST standard for ML-KEM, a key-encapsulation mechanism designed to resist attack by a quantum computer. Finalised in August 2024 alongside FIPS 204 and 205.
- FIPS 204 and FIPS 205
The two post-quantum digital signature standards finalised by NIST in August 2024, published alongside the ML-KEM standard in FIPS 203. Each consumes randomness during key generation and signing.
- FMCFPGA Mezzanine Card
A standard mezzanine connector for attaching modules to an FPGA carrier board, in LPC and HPC variants with an extended FMC+ form. Nellite™ Plus A41 is the Quside entropy module in this format.
G
Top- Gain switching
Driving a laser’s injection current from below its lasing threshold to well above it, once per cycle. Ruby™ N1 does this to erase the phase memory of the previous pulse: with the gain below threshold the cavity field decays away, so the next pulse grows from a fresh seed rather than from what was left behind. It also sets the rate — one sample per cycle.
H
Top- Health test
A test an entropy source runs on itself, at start-up and continuously in operation, to detect failure or degradation. Health testing is what turns a source from a component validated once into one that reports its own condition while it runs.
- HSMHardware security module
A hardened device that generates, stores and uses cryptographic keys inside a controlled boundary. HSMs are among the systems that need reliable seeding and reseeding from a trusted local entropy source.
I
Top- IP licensing
Supplying a design block for integration into a customer’s own silicon, rather than selling a finished chip or module. For an OEM it moves the integration point from the board into the die.
K
Top- Key encapsulation mechanismKEM
A scheme in which one party generates a shared secret and an encapsulation of it that only the holder of the private key can open. ML-KEM, standardised as FIPS 203, is the post-quantum example; encapsulation consumes fresh randomness every time it runs.
L
Top- LOVELocally Verified Entropy
Locally Verified Entropy. Three words describing where the entropy source sits, how its behaviour is established while it runs, and what it delivers.
M
Top- Min-entropy
A worst-case measure of unpredictability, derived from the probability of the single most likely outcome of a source. It is the metric used for cryptographic entropy assessment precisely because the average case is not the interesting case.
- Min-entropy estimation
The process of arriving at a min-entropy figure for a source. Quside estimates against a defined physical and statistical model of the process rather than inferring quality from output statistics alone, because output statistics cannot distinguish a healthy physical source from a well-seeded deterministic one.
- ML-KEM
The lattice-based key-encapsulation mechanism standardised by NIST as FIPS 203 in August 2024. Like every scheme in the family, its security assumes the randomness it consumes is genuinely unpredictable.
N
Top- Nonce
A value used once in a cryptographic protocol or scheme. Nonce reuse or predictability breaks schemes that are otherwise sound, which makes nonces one of the most direct consumers of entropy quality in a system.
O
Top- Operating rangevalidated operating conditions
The conditions — temperature, voltage, supply and age among them — under which an entropy source’s model was validated. A Quside device measures whether it remains inside that range while it runs and reports the result as a health state.
P
Top- PCIePeripheral Component Interconnect Express
The standard expansion bus in servers and appliances. Garnet™ Plus A41 is a low-profile PCIe Gen2 ×4 entropy card delivering up to 1 Gbit/s.
- Phase diffusion
The mechanism behind Ruby™ N1, and the origin of its randomness. The optical phase of a laser field does not hold still: spontaneous emission nudges it at random, and the phase performs a random walk. Ruby™ N1 gain-switches its lasers so that this happens with the cavity depleted, where the diffusion is fastest and no phase memory of the previous pulse survives. Phase itself cannot be measured directly, so the field is interfered with a second one — that turns a phase difference into an intensity a photodiode can read at speed, which mixed-signal electronics digitise into bits.
- PKIPublic key infrastructure
The systems and processes that issue, distribute and revoke certificates and the key pairs behind them. Key generation at scale makes a PKI a heavy and continuous consumer of entropy.
- Post-quantum cryptographyPQC
Cryptographic schemes designed to resist attack by a quantum computer, standardised by NIST in FIPS 203, 204 and 205. PQC algorithms and entropy assurance solve different parts of the same cryptographic architecture: one addresses the algorithm, the other the unpredictability it consumes.
Q
Top- QKDQuantum key distribution
A method of distributing keys between two parties using quantum states, where the security argument rests on physics rather than computational hardness. It is a different technology from quantum random number generation, though QKD systems are themselves consumers of local entropy.
- QRNGQuantum random number generator
A random number generator whose randomness originates in a quantum process. The practical value is not novelty but modelling: a quantum-optical process has a physical model, and a model is something a device can be measured against.
R
Top- Random bit generatorRBG
The complete construction: a validated entropy source feeding an approved deterministic mechanism. SP 800-90C, final in September 2025, describes how the two combine — which is why a certified entropy source is a component of an RBG rather than an RBG in itself.
- Reseeding
Injecting fresh entropy into a deterministic generator that is already running, at defined intervals or after a defined output volume. Appropriate reseeding is one of the conditions under which an approved DRBG delivers what it promises.
- Root of trust
The component in a system whose correctness the rest of the security architecture is anchored to, and which therefore cannot be verified by anything above it. It needs entropy locally and early, before a network is available to supply any.
- Ruby™ N1
Quside’s monolithic photonic entropy chip, built on a dedicated photonic line and delivering up to 1 Gbit/s using quantum phase diffusion. It is the engine inside the Garnet™ PCIe and Nellite™ FMC products.
- Ruby™ S1
Quside’s volume-oriented entropy engine. The output of a VCSEL switches at random between two polarisation states; a polariser between the laser and the photodetector passes one of them and blocks the other, so the detector sees a random sequence of light-or-no-light events. It is the engine inside the compact Onyx™ line.
S
Top- Secure boot
A start-up sequence in which each stage verifies the next before handing control to it, anchored in a root of trust. Secure boot and platform attestation are among the functions that need entropy available locally and early.
- Seeding
Supplying a deterministic generator with entropy at initialisation. The strength of everything the generator subsequently produces is bounded by the quality of what seeded it.
- SP 800-90A
The NIST specification for approved deterministic random bit generator mechanisms — the algorithms that expand a seed into an output stream. It is the deterministic half of the SP 800-90 series.
- SP 800-90B
The NIST specification for entropy sources used in random bit generation, covering the noise source, digitisation, health testing and the estimation of min-entropy. Validation under it produces an entropy certificate; Quside holds E178, scoped to Quside PCIe One and FMC One.
- SP 800-90C
The NIST specification for random bit generator constructions, final in September 2025. It is the document that connects a validated 90B entropy source to an approved 90A deterministic mechanism, making explicit what a certificate on its own does not cover.
- Spontaneous emission
Light emitted by an excited atom or carrier of its own accord, at a random time and with a random phase, rather than in response to an incoming photon. In Ruby™ N1 it is what seeds the random optical phase: while the cavity is depleted below threshold, spontaneous emission leaves a weak field whose phase is unpredictable, and the accumulation of that unpredictability is phase diffusion. The bright pulse that follows is produced by stimulated emission amplifying that seed, which is why the device is not an “amplified spontaneous emission” generator — that is a different QRNG family, one that samples the intensity of incoherent broadband light.
T
Top- TRNGTrue random number generator
A generator whose output derives from a physical process rather than an algorithm. In standards language the corresponding object is the entropy source, which is the term used when validation and scope are being discussed precisely.
V
Top- VCSELVertical-cavity surface-emitting laser
A compact semiconductor laser emitting perpendicular to its wafer surface, manufactured in very high volume for consumer sensing. Its output can switch at random between two polarisation states, and that is the effect Ruby™ S1 uses; Coherent supplies the devices and the material.
Most arguments about randomness are arguments about words.
A certified entropy source is not a random bit generator. A validated min-entropy figure is a measurement of a device in a state, not a permanent property. A scheme category that sounds impressive may be the middle of three. These distinctions decide whether a security argument holds, and they are the first thing an assessor will test.
A term missing, or a definition you would argue with?
If a definition here is wrong, unclear or missing, tell us which one.