packages/knowledge/vertical/aviation-do178c.yaml

# Vertical Governance Policy: Aviation — DO-178C / DO-333 / FAA Order 8110.49
# Doctrine v6 | R3 Adversarial Receipts
# Last revised: 2025-07

schema_version: "1.0.0"
vertical: aviation
regime: DO-178C/DO-333/FAA-8110.49
effective_date: "2025-07-01"
jurisdiction: US-FAA/EASA

meta:
  title: "Aviation AI Governance Policy — DO-178C/DO-333 Alignment"
  description: >
    Maps RTCA DO-178C Software Considerations in Airborne Systems and DO-333
    Formal Methods Supplement to Doctrine v6 Λ-axes for AI/ML components
    in airborne systems, targeting Design Assurance Level (DAL) A–C.
  authority: "RTCA DO-178C (2011); RTCA DO-333 (2011); FAA Order 8110.49B; AC 20-115D; EASA AI Roadmap 2.0"
  receipt_chain_required: true
  merkle_root_algorithm: SHA3-256
  design_assurance_level: DAL-B

regulatory_clauses:
  - clause_id: DO178C-11.9
    title: "Software Accomplishment Summary"
    citation: "RTCA DO-178C § 11.9; FAA AC 20-115D"
    full_ref: "RTCA DO-178C Table A-9 — Software Accomplishment Summary (SAS) lifecycle data"
    lambda_axes:
      - axis: Λ7
        label: Auditability
        weight: 1.0
        enforcement: mandatory
        rationale: >
          AI model version, training data hash, and validation epoch must appear
          in SAS; receipt chain provides cryptographic provenance for each build.
      - axis: Λ1
        label: Transparency
        weight: 0.90
        enforcement: mandatory

  - clause_id: DO178C-6.4
    title: "Software Verification — Structural Coverage"
    citation: "RTCA DO-178C § 6.4; DO-333 § FM.6.4"
    full_ref: "DO-178C § 6.4 — MC/DC structural coverage analysis; DO-333 Supplement for formal methods"
    lambda_axes:
      - axis: Λ8
        label: Robustness
        weight: 1.0
        enforcement: mandatory
        rationale: >
          DAL-B requires MC/DC coverage; AI components must demonstrate
          decision coverage via formal methods or exhaustive test suites;
          coverage receipts committed to Merkle DAG.
      - axis: Λ5
        label: Safety
        weight: 0.98
        enforcement: mandatory

  - clause_id: DO178C-5.5
    title: "Traceability — High-Level Requirements to Software Design"
    citation: "RTCA DO-178C § 5.5; SAE ARP4754B § 5.2"
    full_ref: "DO-178C § 5.5 — Traceability between high-level requirements, low-level requirements, and source code"
    lambda_axes:
      - axis: Λ9
        label: Explainability
        weight: 0.92
        enforcement: mandatory
        rationale: >
          Each AI decision affecting flight-critical functions must be traceable
          to a certified requirement; receipt annotates requirement ID.
      - axis: Λ1
        label: Transparency
        weight: 0.88
        enforcement: mandatory

  - clause_id: DO178C-12.3
    title: "Configuration Management — Change Control"
    citation: "RTCA DO-178C § 12.3; DO-178C Table A-10"
    full_ref: "DO-178C § 12.3 — Problem reporting, change control, change review, configuration status accounting"
    lambda_axes:
      - axis: Λ7
        label: Auditability
        weight: 0.95
        enforcement: mandatory
      - axis: Λ2
        label: Accountability
        weight: 0.88
        enforcement: mandatory
        rationale: >
          Configuration baseline changes to AI model weights require DER-signed
          receipts; change control board approval captured in receipt metadata.

  - clause_id: FAA-8110-49B
    title: "Software Approval Guidelines"
    citation: "FAA Order 8110.49B (2022)"
    full_ref: "FAA Order 8110.49B — Software Approval Guidelines for Airborne Systems"
    lambda_axes:
      - axis: Λ5
        label: Safety
        weight: 1.0
        enforcement: mandatory
        rationale: >
          AI software approval requires Designated Engineering Representative (DER)
          sign-off; each DER attestation is a receipt chain entry with X.509 cert.
      - axis: Λ2
        label: Accountability
        weight: 0.92
        enforcement: mandatory

  - clause_id: EASA-AI-ROADMAP-2
    title: "EASA AI Roadmap — Concept Paper on AI/ML for Aviation Safety"
    citation: "EASA AI Roadmap 2.0 (2023); EASA CP No. 2 (2023)"
    full_ref: "EASA Concept Paper: First usable guidance for Level 1 & 2 ML applications (Oct 2023)"
    lambda_axes:
      - axis: Λ4
        label: Fairness
        weight: 0.72
        enforcement: recommended
        rationale: >
          AI training data must be checked for operational scenario coverage;
          distributional gaps documented in receipt chain data-quality node.
      - axis: Λ9
        label: Explainability
        weight: 0.85
        enforcement: mandatory

  - clause_id: DO333-FM-COMPLETENESS
    title: "Formal Methods — Completeness of Proof"
    citation: "RTCA DO-333 § FM.6.3.2"
    full_ref: "DO-333 § FM.6.3.2 — Completeness of formal proofs with respect to software behavior"
    lambda_axes:
      - axis: Λ8
        label: Robustness
        weight: 0.95
        enforcement: mandatory
        rationale: >
          Formal proofs of AI inference bounds (e.g., Lipschitz constants) must
          be committed as proof-receipts; adversarial perturbation guarantees cited.
      - axis: Λ5
        label: Safety
        weight: 0.95
        enforcement: mandatory

  - clause_id: ARINC-653
    title: "Avionics Application Software Standard Interface"
    citation: "ARINC 653 Part 1-5 (2019)"
    full_ref: "ARINC 653-1 — Avionics Application Software Standard Interface: Partitioning and scheduling"
    lambda_axes:
      - axis: Λ6
        label: Security
        weight: 0.85
        enforcement: mandatory
        rationale: >
          AI inference partitions must be temporally and spatially isolated;
          inter-partition communication events receipt-logged at hypervisor level.
      - axis: Λ10
        label: Sovereignty
        weight: 0.78
        enforcement: recommended

compliance_thresholds:
  minimum_lambda_coverage: 7
  mandatory_axes: [Λ5, Λ7, Λ8]
  receipt_retention_days: 9125        # 25 years per FAA airworthiness
  max_inference_latency_ms: 10        # Safety-critical real-time constraint
  dal_level: B
  coverage_target_percent: 100        # MC/DC for DAL-B

receipt_chain:
  algorithm: SHA3-256
  chaining: merkle_dag
  quorum: 2-of-3
  nodes: [primary, hot-standby, der-archive]
  der_signed: true