Skip to content

Technical Specifications

This appendix is a reference for technical evaluation. Component Specifications gives the specifications of each component, grouped by building block, with the DataHub in a table of its own. Performance Figures collects every quantitative figure in one place and states the basis of each, so that published benchmarks, typical requirements, and design targets are not confused.

Component Specifications

Sign ID (National Digital Identity)
Credentials and identifiers
Credential data model W3C Verifiable Credentials 2.0
Identifiers W3C Decentralized Identifiers (DIDs)
Credential formats VC-JWT, SD-JWT VC, JSON-LD with BBS+
Revocation W3C Bitstring Status List v1.0
Trust registry Ledger-anchored, OpenID Federation 1.0 compatible
Issuer identifiers did:web under government domains, then a government-operated DID method
Credential status Active, suspended, revoked, expired; issuer-signed, cacheable status lists
Transparency log Registry changes, schema versions, status-list digests; no personal data
Cryptography and keys
Signature algorithms ECDSA (secp256k1, P-256), EdDSA (Ed25519), RSA
Zero-knowledge proofs Groth16, Plonk, BBS+ (unlinkable credentials)
Key management HSM-backed (FIPS 140-3 Level 3)
Wallet security Device secure enclave, biometric authentication, encrypted local storage
Holder binding National identity root at onboarding and recovery; device key bound to each presentation
Protocols and interoperability
Issuance OpenID for Verifiable Credential Issuance (OpenID4VCI)
Presentation OpenID for Verifiable Presentations (OpenID4VP)
Offline presentation QR code and NFC
Mobile documents ISO/IEC 18013-5/7 (mDL) compatible
Sign Protocol
Signature schemes ECDSA, EdDSA, RSA
Zero-knowledge proofs Groth16, Plonk, Honk, BBS+
ePassport integration ICAO 9303 compatible

Table 6: Digital identity and credentials specifications

Sign DataHub
Anchoring network Private permissioned network by default, on Hyperledger Besu (QBFT) or Hyperledger Fabric (BFT ordering); a public blockchain after review of every exposed field; anchors recorded as Sign Protocol attestations
Data sharing Common versioned schemas; owner-approved record types for each consumer; access checked by purpose, role, case, and field; exchange receipts
Validators Four in the reference topology (one fault tolerated), expandable to seven (two faults tolerated)
Finality modes Direct on-chain finality; Merkle-batched finality
Cryptography Canonical JSON, SHA-256, Ed25519 or a national scheme, X.509, mTLS, HSM
On-chain data Commitments and minimal technical metadata; no record payloads
Keys Separate keys for the legal decision, the source evidence, and the anchor submission
Verification Source authenticity, evidence integrity, anchor record, network confirmation, and current status; the first four run offline
Recovery Site restored within 4 hours, losing at most 15 minutes of configuration (design target)

Table 7: Sign DataHub specifications

Public blockchain: sovereign smart contracts
Runtime EVM-compatible public Layer 1 network, such as Ethereum
Consensus The host network’s consensus and validator set
Access control Contract-level role-based permissions, address whitelisting, and KYC integration
Upgrades and governance Upgradeable proxy contracts; changes approved through multi-signature wallets held by authorized government entities
Transaction fees Sponsored through contract-level logic or meta-transactions
Token standards ERC-20 for stablecoins; ERC-3643, ERC-1400, ERC-4626, ERC-721, and ERC-1155 for tokenized assets (Real-World Asset Tokenization)
Emergency controls Government-held pause of contract operations
Sign CBDC on Hyperledger Fabric-X
Ledger and tokens
Ledger platform Hyperledger Fabric-X
Token framework Panurus with Fabric Smart Client
Token model UTXO with account-style balance views
Ordering BFT consensus; reference profile of four consenters across primary and disaster recovery sites
Endorsement policy Central bank plus at least one eligible bank
Finality Immediate upon block commitment
Privacy and authorization
Token privacy ZKAT-DLOG without trusted setup (Pedersen commitments, serialized proofs)
Owner authorization Idemix anonymous credentials bound to each transaction challenge
Policy enforcement Mandatory Monetary Auditor gate before ledger submission
Hosting, custody, and keys
Wallet Node hosting Central bank platform, partitioned per bank (recommended), bank-operated, or a combination
Credential custody Hardware-backed TEE or secure enclave in the servicing bank’s domain
Key management HSM-backed certificate authority, TLS, and issuer keys; three-tier PKI
Interfaces
Settlement messaging MT-based RTGS profiles and ISO 20022
Programmability Approved templates; deterministic chaincode for ledger-atomic rules
Interoperability Direct bridge, indirect stablecoin route, interoperability control plane

Table 8: Sovereign money specifications

Public services platform
Channels Mobile app, web portal, and assisted service points
Sign-in Sign ID, with the Sign ID wallet embedded in the app
Services Configurable workflows: forms, eligibility rules, credentials and documents, fees, approvals, deadlines
Payments Sovereign digital money (CBDC or stablecoin) and bank rails
Distribution Batch and scheduled payments; throughput bounded by the underlying ledger
Audit trail Case history for every application; on-chain distribution records
Sovereign AI (Sign AI)
Control layer Sign AI workflow controls: permissions, approvals, case status, audit
Models Government-approved open-weight models; optional external service through a masking gateway, where policy allows
Hosting In-country: government data center, domestic hosting provider, or new national facility; recovery at a second domestic site
Assurance ISO/IEC 27001, ISO/IEC 27701, ISO 22301, and SOC 2 Type II as targets; ISO/IEC 42001 for AI risk management; HSMs certified to FIPS 140-3 Level 3 or Common Criteria
Knowledge Approved, versioned sources with evaluation and rollback
Answer policy Every released answer cites an approved source
Human oversight Authorized review before any action
Decision record Sources and versions, model and rule version, consent, reviewer, final action
Integrations National identity, service portals, government data through the DataHub

Table 9: Public services and sovereign AI specifications

Performance Figures

Component Measure Figure Basis
Credential verification Verification latency P95 under 300 ms Design target
Revocation propagation P95 within 5 minutes Design target
Availability 99.9% at pilot, 99.95% at national scale Design target
Sign DataHub Sustained event intake 2,000 events per second Design target
Burst event intake 5,000 events per second Design target
Acceptance latency P95 under 300 ms Design target
Proof lookup latency P95 under 500 ms Design target
Chain finality Under 10 seconds Design target
Batch root interval 60 seconds or less Design target
Availability 99.9% Design target
Fault tolerance One faulty validator in the four-validator reference topology Reference design
Hyperledger Fabric-X Throughput, one organization on 16 VMs Above 150,000 transactions per second Published benchmark
National retail CBDC Peak throughput Tens of thousands of transactions per second Typical requirement
Confirmation One to two seconds Typical requirement
Availability 99.95% or higher Typical requirement
Fault tolerance One faulty consenter in the four-consenter reference profile Reference design
Sovereign AI compute Pilot configuration 4 GPUs, 1 rack, 6 kW IT capacity Reference design
Production configuration 2 AI servers with 8 data-center GPUs each, 2 racks, 40 kW IT capacity Reference design
Server failure Essential traffic resumes within 5 minutes Design target
Recovery point Up to 15 minutes of data loss Design target
Recovery time Essential service restored within 4 hours Design target

Table 10: Performance figures and their basis

Each figure is labeled with its basis. A published benchmark is reported by the upstream project; a reference design figure follows from the reference topology; a typical requirement is what national programs usually specify; a design target is the figure a deployment is engineered and tested against. The performance of each deployment is established by acceptance testing in its target environment.