ORBIT · Long-horizon evidence

Evidence that outlives its algorithms.

A spacecraft launched now may fly for fifteen years without a patch, and its records may be needed long after that. ORBIT keeps the evidence verifiable as signature suites age: hash-based archive signatures (planned), a renewal ladder run by a named custodian, and checkpoints cosigned by independent witnesses.

Archive structureIn development
RenewalCustodial · named per programme
Zero-knowledge workResearch only
01 The fourth danger

The mission outlives the cryptography it launched with.

Space systems already plan for three dangers: interception, tampering and loss of control. ORBIT adds a fourth: cryptographic survivability. An asset operating for fifteen years or more cannot be patched, and the signature scheme its records depend on may be deprecated mid-life.

The design principle: crypto for the lifetime of the evidence, not of the mission. That means algorithm epochs, two independent assumptions (lattice and hash), hash-based fallbacks (planned: devices will verify LMS signatures, with signer state kept in the ground HSM; SLH-DSA in the archive), periodic re-signing on the ground, independent witnessing and archival verification.

A receipt made today that cannot be verified at an inquiry fifteen years from now makes every other receipt worthless.

02 From chain to vault

Seal, renew, renew again, witness.

Simulated · long-horizon archive01 / 04
01 · SEAL

Receipt chains go into the archive under a manifest.

Receipts are hybrid-signed one by one. Planned: the archive does not re-sign each receipt; it signs checkpoints, roots and archive manifests with SLH-DSA, a stateless hash-based signature, and puts a dual hash (SHA-256 and SHA3-256) on manifests only.

02 · RENEW

Suite B wraps suite A.

Before suite A weakens, a renewal record is issued under suite B. It covers a range of the archive (from hash, to hash, count), names the prior and new suites, cites a timestamp and carries witness cosignatures. The originals are never touched.

03 · RENEW AGAIN

Suite C wraps B and A. A is deprecated, not lost.

Each renewal wraps every older layer. When suite A is deprecated, receipts signed under it still verify because the chain of renewals proves they existed, unchanged, while A was still sound.

04 · WITNESS

Independent witnesses cosign the checkpoints.

Checkpoints follow the C2SP transparency-log checkpoint format and are cosigned by witnesses outside DCS. A verifier rebuilt from the spec checks the whole ladder with DCS servers switched off.

03 The renewal ladder

Each rung wraps the ones below it.

SUITE A

Original seal

Receipts hybrid-signed with Ed25519 and ML-DSA-65. Planned: archive manifests signed with SLH-DSA. The epoch is recorded in every receipt.

SUITE B

First renewal

A renewal record over A's range: what it covers (from hash, to hash, count), the prior suite, the new suite, a timestamp reference and the witness cosignatures.

SUITE C

Next renewal

Wraps B and, through it, A. Triggered by an algorithm-epoch event in the key registry, not by a calendar.

Renewal is custodial. "Verifies for decades" is not a property of any file: renewal must touch every object when a hash weakens, and log lists and trust roots keep rolling. So every ORBIT archive names its custodian and embeds full trust chains; the suite is chosen per use by the programme's approving authority.
04 What the archive is built from

Four mechanisms.

ARCHIVE SIGNATURE

SLH-DSA on manifests, never per receipt

Hash-based signatures are designed for post-quantum security under hash-function assumptions: the conservative second assumption beside lattices. Planned for checkpoints, roots and archive manifests, where size and speed matter least.

Planned

DUAL HASH

SHA-256 and SHA3-256 on manifests

Two hash families on archive manifests only, so a break in one does not silently invalidate the archive. SHA-256 everywhere else.

In development

RENEWAL

Evidence-record renewal

A renewal record in the RFC 4998 evidence-record pattern, pinned to the archive profile so a renewal registered under a different profile is refused at verification.

Built · shape only

WITNESSED CHECKPOINTS

Checkpoints with external cosigners

Checkpoints in the C2SP format with external witness cosignatures. Joining existing witness networks is preferred to running the only witness.

In development

05 A standard that can be written

The formats exist. The post-quantum profile does not.

Every format long-horizon evidence needs already exists: RFC 4998 evidence records, ETSI TS 119 511/512 preservation services, HSS/LMS in COSE and CMS, ML-DSA in COSE, and a recent RFC on state management for stateful hash-based signatures.

The gap is post-quantum evidence records: only individual drafts exist, and long-term preservation services do not yet issue post-quantum timestamps. A post-quantum evidence-record profile is a standard DCS can help write.

The people who need it first are not only in space: preservation providers, public bodies with long retention duties and manufacturers keeping ten-year technical files.

06 Zero-knowledge research

Research only. Never claimed.

REFRAME

A confidentiality overlay on the log

Not "prove the system is safe". Prove a narrow statement without revealing the receipts: "every receipt in this epoch used algorithms from this set", over receipts minted by the acting components.

Future direction

WHICH ZK IS POST-QUANTUM

Hash-based provers, not trusted setups

Pairing-based proofs with a trusted-setup ceremony are classical. Hash-based transparent STARK-style provers are plausibly post-quantum; their cost is proof size, roughly 200–400 KB, not time.

Future direction

THE HARD PART

Completeness

A proof over an inventory the prover assembled is a signed self-attestation. Receipts minted by the acting component, not by the prover, are what close that hole.

Future direction

Any classical zero-knowledge work stays labelled as classical. No post-quantum zero-knowledge claim is made, and on-board zero-knowledge is out of scope.
07 Independence

Two verifiers that agree are not two verifiers.

Byte-identical JavaScript and Python verifiers prove agreement with one conformance corpus, not independence. Real independence means a frozen profile, a published corpus heavy with negative cases, and a verifier that someone who did not write the code can build from the specification alone, without any DCS service.

Independence stepState
JavaScript and Python verifiers, same verdictBuilt
Frozen profilePlanning
Negative-heavy public corpusPlanning
Rebuild from the spec by someone who did not write the codePlanning
08 Questions

On keeping evidence for decades.

Why SLH-DSA in the archive and not everywhere?

Its signatures are large and slow to produce, but it is designed for post-quantum security under hash-function assumptions. That makes it the right second assumption for a few long-lived objects (manifests, roots, checkpoints), not for every receipt.

Are old receipts re-signed when a suite is deprecated?

No. Originals are never altered. A renewal record under the new suite covers them, proving they existed unchanged while the old suite was sound.

Who runs the renewals?

A named custodian. Renewal is a service with an owner, not a property of the file. The custodian and the cosigners are named per programme.

Can the archive be verified without DCS?

That is the goal and the test. Today, offline verification covers signatures, hashes and inclusion; witness cosignatures are still in development, and the archive signature is planned.

Is the zero-knowledge work a product?

No. It is research notes only and is never claimed.

Crypto for the lifetime of the evidence.

Not for the lifetime of the mission, and not for the lifetime of the company that built it.

ORBIT overviewDecisions under link loss