Research libraryThe Industrial Data Spine
Chapter 07 of 14 · Part II · Draft web edition
The data spine must protect production, people and legitimate access at the same time, with controls proportionate to a small team.
The cloth is woven from this chapter's 4 exhibits: each weft is one exhibit's own data, normalised; each warp group is a section. The probe reads the exhibit's own value where it crosses. Procedural visualisation, not a measurement.
Read-out
Derived six-layer protection stack mapping operational technology safety, zero-trust identity, transport security, storage immutability, model governance and tested offline recovery to the industrial data spine.
Unit: control layers. Status: derived. Sources: S22, S24, S25, S26, S0703.
A factory data spine is trustworthy only when its security, privacy and data-access controls are built into the capture boundary, identity model and recovery routine before the first operational feed is connected to analytics.[S22] In GFT's Cameroon textile programme, prepared in January 2024 with source-refresh notes retained in 2026, the sector baseline records about 95 percent of cotton fibre exported raw and about 5 percent processed locally, against a 50 percent local-processing target by 2030 (these inherited entries were not refreshed in 2026 and their original publication dates are unrecorded).[R06] Closing that gap requires factories that capture production, quality, maintenance and energy evidence from the first shift without exposing machinery to unsafe interference, workers to unlawful surveillance, or the enterprise to vendor lock-in.[R07] For a small team without a security department, protection must operate as a compact set of engineering constraints enforced from edge device identity to offline backup restoration.[S24]
When digital projects treat protection as aftercare bolted on after dashboards or machine-learning pilots succeed, three failure modes appear on the shop floor.[S22] First, direct polling or unauthenticated brokers bridged into operational technology (OT) networks can degrade controller scan times, interrupt human-machine interfaces, or expose physical equipment to unauthorised commands.[S22] Second, shared convenience accounts across operators, contractors and ingestion scripts destroy provenance and let one compromised credential alter both historical evidence and live model artefacts.[S25] Third, indiscriminate collection of worker-identifiable productivity or camera records triggers statutory data-protection duties and erodes workforce trust before a lawful basis or human-review safeguard exists.[S29] Treating security, privacy and contractual data rights as primary design constraints prevents these failures within the capacity of a small plant team.[S23]
7.1
Industrial data architecture begins with the physical consequences of failure on the factory floor, not with the convenience of cloud ingestion.[S22] Operational technology covers the programmable logic controllers, drive systems, looms, dyeing controllers, cutting tables and environmental monitors that move matter and energy through a plant.[S22] As the National Institute of Standards and Technology (NIST) sets out in its OT security guidance, control environments differ from corporate IT systems because performance, reliability and physical safety are primary operating requirements.[S22] A delayed packet on an office server causes minor annoyance; a delayed or malformed packet on a dye-house temperature loop or spinning frame can spoil a batch, damage actuators, or endanger operators.[S22] Every component of the data spine touching the shop floor must therefore be evaluated first by its worst-case effect on physical safety and production continuity.[S28]
Standardising the boundary between control execution and enterprise data management stops analytical workloads from destabilising equipment.[S05] Under the IEC 62264 enterprise-control model, real-time sensing, safety interlocks and closed-loop control belong in the lower control domains, while manufacturing operations management, event aggregation and business planning sit across defined interfaces.[S05] The data spine must never reach directly into safety controllers or bypass local interlocks to extract telemetry.[S22] Instead, machine data cross an explicit collection boundary where an edge gateway reads exposed variables at a bounded polling rate or receives outbound publish-subscribe messages without granting external systems write access to control logic.[S06]
Where machinery supports OPC Unified Architecture (OPC UA), the spine should use both its information modelling and its security model to authenticate applications, sign messages and restrict read versus write permissions at the session and node level.[S06] The OPC UA security specification addresses threats across the physical, hardware and software environments in which industrial servers run, including message spoofing, alteration, replay and resource exhaustion from message flooding.[S0703] Where publish-subscribe transport uses MQTT, engineers must recognise that MQTT is a transport protocol rather than a complete security architecture.[S07] Topic namespaces, transport encryption, client certificates and payload validation must be configured on the broker so an analytical subscriber cannot publish commands back to a machine topic or exhaust broker memory.[S07]
As shown in Exhibit 30, protection runs as a continuous chain from the physical machine boundary to the offline recovery vault.[S22] Each layer enforces an engineering constraint so a failure in an upper analytical component cannot propagate downward into production control.[S28] Across this stack, the factory documents two mandatory rules before any data agent is commissioned: no data-collection service, broker or model may become a single point of failure for machine safety or local operation, and uncontrolled experimentation or active network scanning on control networks is prohibited.[S22] What Exhibit 30 cannot establish on its own is the site-specific network topology or firewall rule set, which must be engineered against each plant's physical wiring and controller inventory.[S28]
7.3
Line operator
Line operator: Operational and pipeline write authority 2, Record visibility and personal-data scope 1. Quadrant: Task-specific line capture · illustrative
In a small factory, access control often fails not because encryption is weak, but because convenience encourages shared accounts and over-privileged service tokens.
Shift supervisor
Shift supervisor: Operational and pipeline write authority 4, Record visibility and personal-data scope 4. Quadrant: Task-specific line capture · illustrative
Quality steward
Quality steward: Operational and pipeline write authority 5, Record visibility and personal-data scope 3. Quadrant: Automated technical custody · illustrative
Maintenance engineer
Maintenance engineer: Operational and pipeline write authority 4, Record visibility and personal-data scope 2. Quadrant: Task-specific line capture · illustrative
Data analyst
Data analyst: Operational and pipeline write authority 2, Record visibility and personal-data scope 5. Quadrant: Scoped oversight and audit · illustrative
Equipment vendor
Equipment vendor: Operational and pipeline write authority 3, Record visibility and personal-data scope 1. Quadrant: Task-specific line capture · illustrative
Deployment service
Deployment service: Operational and pipeline write authority 8, Record visibility and personal-data scope 2. Quadrant: Automated technical custody · illustrative
Illustrative permission mapping across seven factory roles on two governance dimensions: x measures operational and model write authority (0 = read/append local only, 10 = full pipeline or infrastructure write), y measures scope of sensitive and identifiable record visibility (0 = anonymised or single-asset only, 10 = plant-wide or personal-data review). Positions are illustrative design targets based on S25, S29 and S0701, not empirical measurements.
Unit: illustrative score (0 to 10). Status: illustrative. Sources: S25, S29, S0701. Illustrative figures are teaching examples, not measured factory results.
7.3 · the bodyExhibit 31 above; its figures are illustrative where marked.
In a small factory, access control often fails not because encryption is weak, but because convenience encourages shared accounts and over-privileged service tokens.[S25] When supervisors share a terminal password, or when an ingestion script uses the database administrator's master credentials, the spine loses both security and provenance.[S09] If a KPI threshold is altered or a batch record is corrected, the audit log cannot show which person or automated job made the change.[S02] Applying least privilege under NIST security and privacy controls requires granting each human user and automated process only the minimum authorisations needed for its assigned task.[S25]
Modern industrial architecture reinforces least privilege with zero-trust principles: no device, user account or service receives implicit trust merely because it sits inside the local area network or on a company workstation.[S0701] Authentication and authorisation of both subject and device occur as discrete checks before a session is established to any data resource.[S0701] This requires separating human identities from machine identities across every stage of the spine.[S25]
For human personnel, permissions follow the governance roles established in this paper: line operators record and view shift events for their work centre; domain stewards review quarantine queues and submit correction events; technical custodians manage storage, backups and network conduits without authority to alter business definitions; and domain owners approve schema versions, KPI formulas and access exceptions.[S02] External equipment vendors never receive standing administrative credentials; their access is provisioned only for approved maintenance windows through an audited jump host and revoked immediately afterward.[S24]
For non-human processes, every edge gateway, MQTT publisher, ingestion worker, transformation job, training pipeline and registry writer receives its own cryptographic identity and narrowly scoped permission set.[S15] An edge collector on spinning line A needs permission to publish signed telemetry to the raw topic for line A; it has no reason to read order margins, query attendance records, or write to the curated KPI store.[S25] Similarly, a training pipeline needs read-only access to frozen dataset snapshots and write access to the candidate experiment log, never credentials that overwrite the production model pointer or mutate raw events.[S15] Narrowly scoping machine credentials also limits the blast radius of adversarial machine-learning threats such as training-data poisoning or model tampering across the lifecycle stages defined in NIST's adversarial machine-learning taxonomy.[S0702]
The permission separation in Exhibit 31 shows why no single role should combine broad pipeline write authority with unrestricted visibility into identifiable worker or commercial records.[S25] In this illustrative matrix, a deployment service holds high write authority to promote approved model containers (x = 8 on the illustrative zero-to-ten scale), yet its visibility into personal records stays near the floor (y = 2) because it processes only governed operational features.[S29] Conversely, a data analyst inspects cross-line curated views (y = 5 on the illustrative scale) while holding minimal write authority (x = 2), preventing silent manual edits to production tables.[S25] What Exhibit 31 cannot replace is a concrete access-control list in the identity provider and database engine, reviewed whenever staff change role or leave the factory.[S25]
…approved model containers (x = 8 on the illustrative zero-to-ten scale)
S29 · illustrative
To handle genuine emergencies without weakening daily controls, the spine includes a documented break-glass procedure.[S25] If an urgent night-shift fault requires elevated access when the primary custodian is unavailable, a time-locked emergency credential can be activated by the duty manager.[S27] Activation triggers an audit alert, expires after an illustrative four-hour maintenance window, and requires a formal review at the next morning meeting to verify every command executed.[S25]
7.5
Small manufacturers rarely possess a security operations centre, and copying the control catalogue of a bank guarantees shelfware that nobody follows.[S24] The Cybersecurity and Infrastructure Security Agency (CISA) designed its Cross-Sector Cybersecurity Performance Goals to help small and medium organisations establish a prioritised starting posture.[S24] Combined with the governance function in the NIST Cybersecurity Framework 2.0, which embeds cybersecurity accountability into leadership and enterprise risk management, a small plant can maintain a durable baseline without a specialist department.[S23] Rather than treating these references as one universal checklist, GFT adapts them into six hygiene disciplines within the ISA/IEC 62443 industrial control lifecycle.[S28]
Combined with the governance function in the NIST Cybersecurity Framework 2.0
S23
0 permanent unmonitored remote connections as an illustrative target), using time-bounded jump-host sessions approved by the maintenance owner and logged end to end.[S24]Method
System boundary and verification checks for the small-team security baseline
7.6
A green checkmark on a nightly backup scheduler does not prove that a factory can recover its data spine after ransomware, hardware failure or operator error.[S26] Joint guidance from CISA, MS-ISAC, NSA and the FBI emphasises two requirements: maintain offline, encrypted backups of critical data, and regularly test their availability and integrity in a disaster-recovery scenario.[S26] Backing up only the raw sensor table is insufficient; if schemas, master-data hierarchies, KPI dictionary versions, transformation scripts, provenance graphs and approved model artefacts are lost, the restored numbers are uninterpretable.[S15]
The recovery scope therefore encompasses the full semantic and operational chain: the append-only raw event store, versioned master data, the KPI dictionary and contracts, PROV-O lineage records, model registry metadata and deployment configurations.[S09] At least one encrypted copy is held offline or in an immutable vault whose retention lock cannot be shortened or deleted using the credentials of the active database administrator or pipeline service.[S26]
Proof of recoverability comes from restoration drills into an isolated environment, never from backup logs alone.[S26] During a drill, the custodian and a domain steward take the offline backup media, provision a clean isolated host, restore the raw records and metadata, recompute a historical shift KPI sheet and re-run a historical model batch.[S15] The test passes only when actual recovery time meets the recovery time objective (RTO), the missing-data interval meets the recovery point objective (RPO), and restored KPI outputs match the signed management report.[S26]
Interlude · the cloth half-woven
Woven: sections 7.0 to 7.6Still to weave: 7.7 to 7.11
Read-out
Inherited GFT programme evidence base prepared 31 January 2024. R06 and R07 were not refreshed in 2026 and their original publication dates are unrecorded. Figures describe sector-level baseline and policy ambition, not a current GFT factory measurement or causal productivity claim.
Unit: percent of cotton fibre. Status: sourced. Sources: R06, R07.
Read this exhibit in its chapter ↗7.7
Detect and triage
0: Detect and triage. Broker rate alert or unauthorized write attempt triggers steward and custodian paging. · illustrative
When an anomaly strikes a connected factory, ambiguity over who has authority to disconnect a network conduit or suspend an automated model costs more time than the technical fault itself.
Isolate conduit
15: Isolate conduit. Custodian and OT lead sever the OT-to-IT firewall conduit while edge buffers continue local capture. · illustrative
Switch to manual fallback
30: Switch to manual fallback. Production manager activates paper/local terminal logging and suspends model-assisted recommendations. · illustrative
Preserve forensic evidence
60: Preserve forensic evidence. Custodian snapshots volatile memory, authentication logs and quarantined payloads to read-only media. · illustrative
Eradicate and restore
120: Eradicate and restore. Compromised credentials rotated; clean environment restored from offline encrypted backup. · illustrative
Reconcile and reconnect
180: Reconcile and reconnect. Edge queues replayed idempotently; steward verifies KPI parity and releases quarantine. · illustrative
Close and update runbook
240: Close and update runbook. Cross-functional review records root cause, missing controls and updated escalation thresholds. · illustrative
Illustrative seven-step incident-response and recovery sequence measured in minutes from initial anomaly detection (minute 0 to minute 240). Timings are illustrative drill targets derived from the procedural stages in S22, S26 and S27, not empirical benchmarks.
Unit: elapsed minutes from detection (illustrative). Status: illustrative. Sources: S22, S26, S27. Illustrative figures are teaching examples, not measured factory results.
7.7 · the bodyExhibit 32 above; its figures are illustrative where marked.
When an anomaly strikes a connected factory, ambiguity over who has authority to disconnect a network conduit or suspend an automated model costs more time than the technical fault itself.[S27] If an IT administrator hesitates to isolate a compromised broker for fear of halting a weaving room, or if a line supervisor continues following corrupted model recommendations because no manual fallback rule was issued, a containable incident escalates into an operational shutdown or quality recall.[S22] NIST's incident-response guidance aligned with the Cybersecurity Framework 2.0 stresses that response procedures must be tested and exercised periodically both to verify their accuracy and to train personnel before a crisis occurs.[S27]
…aligned with the Cybersecurity Framework 2.0 stresses that response procedures must be tested and…
S27
In a manufacturing environment, incident response is not an abstract tabletop discussion for software engineers; it is a physical shop-floor drill comparable to a fire drill or a lockout-tagout safety exercise.[S22] The response plan must predefine clear decision authority across five roles: the production manager (who owns physical line operation and manual fallback execution), the OT/maintenance lead (who owns controller safety and physical network disconnection), the data custodian (who owns broker isolation, evidence preservation and backup restoration), the domain steward (who owns data quarantine and post-recovery reconciliation), and the plant director (who owns external legal, regulatory, customer and vendor notifications).[S23]
The sequence in Exhibit 32 shows how containment and production continuity run in parallel.[S22] Within an illustrative 15 minutes of confirming an unauthorised write or broker anomaly, the OT lead and data custodian isolate the firewall conduit between the shop floor and the server room.[S22] Because the edge architecture is engineered for disconnection, machines remain running safely under local loop control and operators switch by minute 30 (illustrative) to rehearsed fallback sheets or local terminal logging.[S33] Forensic evidence is captured at minute 60 before any system is wiped, clean restoration from offline backups occurs at minute 120, and idempotent replay reconciles the buffered interval by minute 180, followed by review at minute 240.[S26] While the minute marks in Exhibit 32 are illustrative drill targets, the ordering is mandatory: isolate the control boundary first, stabilise safe manual production second, preserve evidence third, and reconnect only after clean restoration and steward reconciliation.[S27]
Within an illustrative 15 minutes of confirming an unauthorised write or broker anomaly
S22 · illustrative
Rehearsing this sequence twice a year exposes hidden dependencies that diagrams miss, such as a fallback sheet lacking the current lot-identifier format, a backup decryption passphrase locked in an absent manager's desk, or an inspection station whose lighting controller depended on the office network switch.[S27] Every defect discovered during a drill enters the governance action register with a named owner and due date.[S02]
7.8
A factory data spine captures events generated by machines and human operators working together, so operational telemetry can easily cross into personal-data processing.[S29] Attendance logs, operator-tagged sewing outputs, maintenance sign-offs, skill matrices and overhead cameras aimed at inspection tables can all identify individual workers directly or indirectly.[S29] In Cameroon, where the inherited register notes that informal employment accounts for about 90 percent of the labour market alongside a standard 40-hour workweek and 48-hour maximum with overtime at 1.2 times normal pay (inherited entry not refreshed in 2026; original publication date unrecorded), formal industrial employers carry a clear duty to show that digitalisation respects labour standards and worker dignity.[R16]
…informal employment accounts for about 90 percent of the labour market alongside a standard…
R16
Two principles govern worker-related information in the spine: strict data minimisation at collection, and mandatory human authority over employment decisions.[S29] Under Recital 39 of the European Union General Data Protection Regulation (GDPR), personal data must be adequate, relevant and limited to what is necessary for the stated purpose.[S29] On the shop floor, minimisation means the spine defaults to asset-level, lot-level and shift-level identifiers rather than person-level identifiers wherever the operational decision does not require knowing the individual worker.[S29]
Under Recital 39 of the European Union General Data Protection Regulation (GDPR)
S29
To calculate line overall equipment effectiveness (OEE), track cotton-lot genealogy, balance work-in-progress buffers, or allocate electricity and steam per kilogram of dyed fabric, the spine needs the machine ID, line ID, style code, order number, shift code and event timestamp.[S05] None of those calculations requires a worker ID on every piece-count message or continuous video of operators in a central archive.[S29] Where cameras support automated fabric-defect detection, the optical field of view should be framed mechanically onto the moving textile web and inspection platen, excluding workers' faces and bodies at the lens boundary rather than relying on downstream blurring.[S29]
Where a legitimate purpose requires identifiable worker records, such as statutory payroll and overtime compliance, safety incident reporting, or certified operator qualification for hazardous dyeing and boiler plant, those records belong in a restricted people-domain store with separate access controls and shorter retention schedules.[S25] Analytical views published to engineers or machine-learning pipelines strip direct identifiers and aggregate metrics to the team, line or shift grain so operational analytics cannot become covert surveillance.[S29]
Equally important is the boundary on automated workplace decisions.[S30] Principle 5.5 of the International Labour Office (ILO) code of practice on workers' personal data states that decisions concerning a worker should not be based solely on automated processing of that worker's personal data.[S30] Although the ILO code is non-binding guidance rather than statute, GFT adopts Principle 5.5 as a hard constraint across all factory analytics and MLOps workflows.[S30] No KPI score, vision defect attribution or model output may automatically trigger discipline, pay deduction, bonus forfeiture, demotion or dismissal.[S30] Any workforce-related indicator serves at most as an inquiry prompt for a trained supervisor, who must inspect machine conditions, material quality and tooling before reaching a judgement under the human-oversight principles of NIST's AI Risk Management Framework and ISO/IEC 23894.[S11]
7.10
Beyond engineering minimisation and internal rules, a factory in Cameroon must comply with national data-protection law.[S31] On 23 December 2024, the Presidency of the Republic of Cameroon enacted Law No. 2024/017 relating to personal data protection in Cameroon.[S31] That statute regulates the collection, recording, storage, modification, transfer and destruction of personal data within the national territory.[S31] Crucially for industrial digitalisation, Section 19(1) of Law No. 2024/017 states verbatim: "The processing of personal data shall be subject to prior authorization by the Personal Data Protection Authority."[S31]
Section 19(1) has immediate consequences for how the data spine is phased and documented.[S31] An engineering team cannot switch on biometric time-clocks, operator-linked productivity tracking, cloud human-resources analytics or cross-border replication of identifiable employee logs and assume internal policy notes satisfy the law.[S31] Before any identifiable personal data are collected, linked across domains or transmitted outside Cameroon, the enterprise must obtain current local legal advice from qualified Cameroonian counsel on prior-authorisation filings, applicable exemptions, the operational status and procedures of the Personal Data Protection Authority, and cross-border transfer rules.[S31] This white paper provides engineering design guidance only; it is not legal, regulatory, investment or tax advice.[S31]
Section 19(1) has immediate consequences for how the data spine is phased and documented
S31
Read-out
| Machine cycle and energy logs | OEE, maintenance and carbon intensity | Non-personal (asset and shift ID only) | Permitted for anomaly and drift alerts | Verify vendor contract rights (S32 where applicable) |
|---|---|---|---|---|
| Fabric roll vision images | Automated defect detection and grading | Non-personal (camera framed on fabric web) | Operator-assisted roll disposition | Physical lens check; local edge storage default |
| Order and lot genealogy | Customer traceability and rework control | Non-personal (lot, order and station ID) | Rule-based hold and quarantine routing | Commercial confidentiality; export format contract |
| Operator attendance and payroll | Statutory wage, overtime and shift safety | Personal data (named worker identity) | No automated discipline or pay deduction (S30) | Law No. 2024/017 Sec. 19(1) review before processing |
| Safety and machine qualification | Verify certified operator for hazardous plant | Personal data (worker ID and licence date) | Hard interlock check only; human sign-off | Local counsel authorisation review; local hosting |
| Custodian and break-glass audit logs | Security forensics and privilege accountability | Personal data (administrator account ID) | Alert routing to security owner only | Defined retention window; restricted vault access |
Derived processing-register structure separating non-personal machine and lot records from identifiable worker records across purpose, identifiability, access, retention, automated-decision limits, Cameroonian legal screening and cross-border routing.
Unit: governance register schema. Status: derived. Sources: S29, S30, S31, S32.
Maintaining the processing register in Exhibit 33 gives management, legal counsel and workforce representatives one inspectable inventory of every dataset in the factory.[S29] Separating non-personal machine, energy, vision and lot records from the narrow set of personal records required for payroll, safety certification and security auditing reduces the statutory compliance surface while protecting worker privacy by design.[S29] For every row classified as personal data, Exhibit 33 binds the dataset to the ILO non-automation safeguard and holds collection or reuse behind an explicit legal-review gate under Cameroon's Law No. 2024/017.[S31] What Exhibit 33 cannot do is replace a formal legal opinion or regulatory filing with the Cameroonian authority, which must be executed by local counsel before deployment.[S31]
7.11
Related · chapter 086 paragraphs step it
Read-out
Derived portability assessment across eight required export artefact classes, combining enterprise-control models (S05), OPC UA information models (S06), EPCIS event semantics (S08), PROV-O lineage (S09), MLOps pipeline artefacts (S15) and connected-product data access principles (S32). X-axis scores semantic lock-in exposure if left unstandardised (0=low, 10=high); Y-axis scores independent restoration test complexity required before accepting an export routine (0=low, 10=high).
Unit: derived index (0 to 10). Status: derived. Sources: S05, S06, S08, S09, S15, S32.
Read this exhibit in its chapter ↗The final protection constraint covers commercial and legal control over operational data generated by purchased machinery and software.[S32] Emerging-market textile plants acquire spinning frames, looms, dyeing vessels, compressors and inspection rigs from international original equipment manufacturers (OEMs).[S06] Modern machinery frequently ships with embedded sensors and proprietary gateways that transmit telemetry to the vendor's cloud portal while locking the factory out of the raw stream, charging recurring fees to export the plant's own operating history, or barring the factory from sharing diagnostics with an independent maintenance or analytics team.[S32]
When an OEM contract restricts access to machine telemetry, the data spine is crippled at capture.[S05] The plant cannot join machine states to its order genealogy, audit how the vendor's dashboard calculates availability, or migrate to a new analytics tool without losing history.[S09] Within the European Union, Regulation (EU) 2023/2854 (the EU Data Act) addresses this directly: Recital 5 states that users of a connected product or related service in the Union can access, in a timely manner, the data generated by the use of that product or service and share them with third parties of their choice.[S32]
For a factory in Cameroon or structured through Hong Kong SAR, the EU Data Act does not apply automatically by territory; its obligations depend on whether the user, product or service falls within the Regulation's territorial and material scope.[S32] An engineering team therefore never assumes foreign statutory rights will protect a Cameroonian mill automatically.[S32] Instead, GFT translates those access principles and industrial interoperability standards into five procurement clauses negotiated before any machinery or software order is signed:[S06]
Designing security, privacy and contractual data rights into the spine from the outset prevents the delays caused by safety trips, ransomware lockouts, labour disputes, regulatory stop-orders and vendor lock-in.[S23] With the protection stack, access matrix, incident-response drill, processing register and procurement clauses in place, the factory can now address the next requirement of Part II: keeping the spine continuous during network outages and portable across vendor changes.[S33]
Check the record
Every knot is a source this chapter cites; select one to read its record, or filter the register by grade. Publication dates and inherited-source limits stay explicit. Read the paper's full evidence summary.
26 sources · 172 citation marks on this pageMost cited: S22, 23 times
Grade AGrade BGrade CKnot size: citations on this page
All 26 records shown
R06 · Grade C · inherited record
Inherited evidence R06
About 95 percent of cotton fibre exported raw against 5 percent processed locally; industrialisation target of 50 percent local processing.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R07 · Grade C · inherited record
Inherited evidence R07
SND30 restates the 50 percent local-processing target by 2030.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
R16 · Grade C · inherited record
Inherited evidence R16
Informal share of employment about 90 percent; standard 40-hour week, 48-hour maximum, overtime at 1.2 times.
Publication date unrecorded. Inherited source-text record. The original publication has not been independently reopened for this web edition.
c07-S02 · Grade A
ISO 8000-150:2022: Data quality, Part 150: Data quality management: Roles and responsibilities
This document specifies the key considerations for organizations that are establishing appropriate roles and responsibilities for data quality management.
2022. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S05 · Grade A
IEC 62264-1:2013, Edition 2.0: Enterprise-control system integration, Part 1: Models and terminology
This part of IEC 62264 provides standard models and terminology for describing the interfaces between the business systems of an enterprise and its manufacturing operations and control systems.
2013. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S06 · Grade A
OPC Unified Architecture, Part 1: Overview and Concepts, OPC 10000-1, version 1.05.06
It provides a comprehensive framework for modelling information, secure communication, and ensuring interoperability between diverse systems and devices.
2025-10-22. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S07 · Grade A
MQTT Version 5.0
MQTT is a Client Server publish/subscribe messaging transport protocol.
2019-03-07. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S09 · Grade A
PROV-O: The PROV Ontology
It provides a set of classes, properties, and restrictions that can be used to represent and interchange provenance information generated in different systems and under different contexts.
2013-04-30. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S11 · Grade A
Artificial Intelligence Risk Management Framework, AI RMF 1.0, NIST AI 100-1
Human judgment should be employed when deciding on the specific metrics related to AI trustworthiness characteristics and the precise threshold values for those metrics.
2023-01-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S14 · Grade A
ISO/IEC 23894:2023: Information technology, Artificial intelligence, Guidance on risk management
The guidance also aims to assist organizations to integrate risk management into their AI-related activities and functions.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S15 · Grade B
MLOps: Continuous delivery and automation pipelines in machine learning
Information about each execution of the ML pipeline is recorded in order to help with data and artifacts lineage, reproducibility, and comparisons.
2024-08-28. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S22 · Grade A
Guide to Operational Technology Security, NIST SP 800-82 Revision 3
This document provides guidance on how to secure operational technology (OT) while addressing their unique performance, reliability, and safety requirements.
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S23 · Grade A
The NIST Cybersecurity Framework 2.0
Governance activities are critical for incorporating cybersecurity into an organization's broader enterprise risk management (ERM) strategy.
2024-02-26. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S24 · Grade A
Cross-Sector Cybersecurity Performance Goals, Version 2.0
They aim to help critical infrastructure entities, particularly small and medium organizations, get started on their path toward a strong cybersecurity posture.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S25 · Grade A
Security and Privacy Controls for Information Systems and Organizations, NIST SP 800-53 Revision 5
Employ the principle of least privilege, allowing only authorized accesses for users (or processes acting on behalf of users) that are necessary to accomplish assigned organizational tasks.
2020-12-10. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S26 · Grade A
#StopRansomware Guide
Maintain offline, encrypted backups of critical data, and regularly test the availability and integrity of backups in a disaster recovery scenario [CPG 2.R].
2023. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S27 · Grade A
Incident Response Recommendations and Considerations for Cybersecurity Risk Management: A CSF 2.0 Community Profile, NIST SP 800-61 Revision 3
Procedures can be tested or exercised periodically to verify their accuracy and can be used to help train new personnel.
2025. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S28 · Grade A
ISA/IEC 62443 Series of Standards
The ISA/IEC 62443 series addresses the security of industrial automation and control systems (IACS) throughout their lifecycle (which applies to all automation and control systems, not only industrial).
2026. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S29 · Grade A
Regulation (EU) 2016/679, General Data Protection Regulation
The personal data should be adequate, relevant and limited to what is necessary for the purposes for which they are processed.
2016-04-27. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S30 · Grade A
Protection of workers' personal data: An ILO code of practice
Decisions concerning a worker should not be based solely on the automated processing of that worker's personal data.
1997. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S31 · Grade A
Law No. 2024/017 of 23 December 2024 relating to personal data protection in Cameroon
The processing of personal data shall be subject to prior authorization by the Personal Data Protection Authority.
2024-12-23. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S32 · Grade A
Regulation (EU) 2023/2854 on harmonised rules on fair access to and use of data, Data Act
This Regulation ensures that users of a connected product or related service in the Union can access, in a timely manner, the data generated by the use of that connected product or related service and that those users can use the data, including by sharing them with third parties of their choice.
2023-12-13. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S33 · Grade B
Understand extended offline capabilities for IoT Edge devices, modules, and child devices
If your devices go offline, the IoT Edge parent device stores all device-to-cloud messages until the connection is reestablished.
2024-06-10. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S0701 · Grade A
Zero Trust Architecture, NIST SP 800-207
Zero trust assumes there is no implicit trust granted to assets or user accounts based solely on their physical or network location (i.e., local area networks versus the internet) or based on asset ownership (enterprise or personally owned).
2020-08. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S0702 · Grade A
Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations, NIST AI 100-2 E2023
The taxonomy is built on surveying the AML literature and is arranged in a conceptual hierarchy that includes key types of ML methods and lifecycle stages of attack, attacker goals and objectives, and attacker capabilities and knowledge of the learning process.
2024-01. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗c07-S0703 · Grade A
OPC Unified Architecture, Part 2: Security Model, OPC 10000-2, version 1.05.06
This part describes the OPC Unified Architecture (OPC UA) security model. It describes the security threats of the physical, hardware, and software environments in which OPC UA is expected to run.
2025-10-22. Register access date: 2026-10-30. Source provenance is supplied by the paper evidence register.
Open the recorded source ↗S22 National Institute of Standards and Technology. 2023.
R06 Cameroon Trade Hub. Publication date unrecorded; inherited source-text check.
R07 Ecofin Agency. Publication date unrecorded; inherited source-text check.
S24 Cybersecurity and Infrastructure Security Agency. 2025.
S25 National Institute of Standards and Technology. 2020-12-10.
S29 European Parliament and Council of the European Union. 2016-04-27.
S23 National Institute of Standards and Technology. 2024-02-26.
S28 International Society of Automation. 2026.
S05 International Electrotechnical Commission. 2013.
S06 OPC Foundation. 2025-10-22.
S0703 OPC Foundation. 2025-10-22.
S07 OASIS Open. 2019-03-07.
S26 CISA, MS-ISAC, NSA and FBI. 2023.
S09 World Wide Web Consortium. 2013-04-30.
S02 International Organization for Standardization. 2022.
S0701 National Institute of Standards and Technology. 2020-08.
S15 Google Cloud. 2024-08-28.
S0702 National Institute of Standards and Technology. 2024-01.
S27 National Institute of Standards and Technology. 2025.
S33 Microsoft. 2024-06-10.
R16 WageIndicator. Publication date unrecorded; inherited source-text check.
S30 International Labour Office. 1997.
S11 National Institute of Standards and Technology. 2023-01-26.
S31 Presidency of the Republic of Cameroon. 2024-12-23.
S32 European Parliament and Council of the European Union. 2023-12-13.
S14 ISO/IEC. 2023.
Keep the thread · chapter 07 of 14
Move the shuttle along the rack to preview any chapter: its sections, exhibits and reading time. Draft web edition. This research is not investment, legal or tax advice; illustrative models are not measured factory results.
The paper's rack · 14 boltsBolt length: reading time
01Part I: Govern meaning before scale
The binding constraint on industrial AI is often the broken chain between work, record and decision, not the absence of an algorithm.
02Part I: Govern meaning before scale
A workable governance model fits into existing factory roles and makes every definition, exception and release somebody's explicit decision.
03Part I: Govern meaning before scale
Shared identifiers, event semantics and KPI formulas let unlike systems describe the same work without hiding their differences.
04Part I: Govern meaning before scale
A small set of enforced checks, visible exceptions and complete lineage is more useful than an unmeasured claim that the data are clean.
05Part II: Build the minimum durable spine
A durable architecture preserves source evidence, adds governed context once and lets each decision consume a controlled product.
06Part II: Build the minimum durable spine
A value becomes useful evidence only when the factory can establish what it describes, when it happened and which state or transformation it represents.
07Part II: Build the minimum durable spine
The data spine must protect production, people and legitimate access at the same time, with controls proportionate to a small team.
08Part II: Build the minimum durable spine
A factory proves resilience by operating through disconnection, restoring its records and leaving a vendor without losing its history.
09Part III: Control models as production change
The first model gate is operational and economic: nobody should train before the action, baseline, data rights and stop conditions are explicit.
10Part III: Control models as production change
A model becomes a release candidate only when another competent person can reconstruct what data, code, configuration and evaluation produced it.
11Part III: Control models as production change
Release authority follows demonstrated local fitness, safe fallback and accountable acceptance, not a strong aggregate validation score.
12Part III: Control models as production change
A useful monitor detects change, establishes whether it matters and routes a named person through a rehearsed response.
13Part IV: Prove value and scale deliberately
The spine earns further investment when people use it, decisions improve and net operational benefit survives a comparison with the baseline.
14Part IV: Prove value and scale deliberately
GFT should scale the spine only as ownership, meaning, recovery, reproducibility and net value are demonstrated in production.