Network Digital-Twin Fidelity Director

The mandate

The operator built a digital twin to evaluate fibre maintenance, routing changes and capacity additions, but several approved scenarios produced real convergence, optical margin and customer effects materially worse than forecast. Topology data is accurate at inventory level while operational state, protection behaviour and service mappings lag. The defined problem is to determine which decisions the twin can support and what fidelity evidence each requires.

The named deliverable is a Network Digital-Twin Fidelity Standard and Governed Planning Architecture. It must define decision classes, source authority, temporal resolution, optical and packet models, service mapping, calibration, uncertainty, validation datasets, scenario controls, versioning, operator workflow, economic use and prohibited conclusions, accompanied by a tested reference scenario library.

Four milestones govern four months: by week two, accept the decision and source inventory; by week six, deliver model-gap and lineage findings; by week twelve, complete calibration against historical changes and controlled field observations; and by week seventeen, submit the fidelity standard, reference library and adoption plan. Invoices follow acceptance of each milestone artefact.

The planning officer and investment group will accept the work only when at least ten past events replay within agreed envelopes for route, optical and service outcomes, uncertainty is visible in every scenario, source freshness is enforced, planners can reproduce three reference decisions without the consultant, and the standard blocks use of the twin where evidence is insufficient.

The client will provide topology and service inventories, optical measurements, routing events, maintenance history, customer incidents, GIS data, planning assumptions, economic cases, field access and named engineers. The consultant will not operate the live network, approve maintenance, procure modelling software, redesign customer products or guarantee forecast accuracy outside validated decision classes; unavailable histories remain declared limitations.

Why this is external work

The internal team built the twin and is measured partly by its adoption, while planning groups already cite its outputs in contested investments. Vendors can validate their own modelling components but not the complete operational representation. Independent network and simulation judgment is required to prevent a visually persuasive model from acquiring decision authority beyond its demonstrated fidelity.

What you will own

• Classify planning decisions by required topology, optical, routing, traffic, service, operational and economic fidelity.
• Trace source authority, timestamp, transformation and uncertainty for inventory, measurements, events, service mappings and assumptions.
• Reproduce historical maintenance, failure, convergence and capacity events to quantify model error and missing mechanisms.
• Define calibration and validation envelopes for optical margin, path choice, convergence, utilisation, service impact and restoration.
• Build controlled scenario libraries with versioned inputs, expected outputs, confidence, prohibited extrapolation and independent review.
• Specify planner workflow, change governance, monitoring and retirement when live conditions exceed a validated model boundary.
• Deliver the standard, governed architecture, reference scenarios, evidence corpus and prioritised model-improvement backlog.

Candidate qualifications

• Built or independently validated production network digital twins spanning optical, packet and customer-service layers.
• Calibrated simulations against real changes and failures, exposing mechanisms hidden by accurate-looking topology.
• Defined uncertainty and permitted-use boundaries so decision makers could not overinterpret model output.
• Reconciled inventory, telemetry, GIS, routing, optical and service data with differing freshness and ownership.
• Used field observation and historical replay to reject or constrain a network investment scenario.
• Delivered a governed simulation method that planners operated and challenged after the consultant departed.

Non-negotiables

• Can complete three Swedish operations and field-verification visits inside the four-month milestone calendar.
• Will disclose relationships with network-modelling, GIS, optical, routing and planning-tool suppliers.
• Brings multi-layer network simulation validation; dashboard digital twins or static topology modelling alone are insufficient.
• Accepts reproducible historical replay and explicit prohibited-use rules as final acceptance requirements.

1. 49 words maximum. Describe a network twin with accurate topology that still predicted the wrong operational outcome.
2. 49 words maximum. Which historical event would you select first to calibrate optical and routing interaction?
3. 49 words maximum. How would you stop planners using a high-fidelity model outside its validated decision class?