One Size Does Not Secure All Why Critical Infrastructure Needs Customisable OT Security
How sector-aware detection, asset-specific context and configurable use cases make OT security more effective across diverse critical infrastructure environments.
Velos Shield Case Study Series
VELOSSHIELD CASE STUDY SERIES
Executive Summary
Critical infrastructure organisations share a need for reliable OT cybersecurity, but they do not share identical operational realities. An airport, power utility, water treatment plant, refinery, railway network and manufacturing facility may all use industrial control systems—yet their critical assets, operating constraints, communication patterns, safety consequences and adversary exposure can differ substantially. A generic OT security tool can provide useful asset discovery and protocol visibility. Its effectiveness can decline, however, when it treats every environment as if the same assets, behaviours and threat scenarios matter equally. A detection that is meaningful in one sector may be routine in another; an apparently minor deviation in a specialised control process may be highly significant in its local context. This case study explores the operational case for a highly customisable OT security capability: one that allows defenders to adapt asset models, protocol parsing, detection logic, risk weighting, threat scenarios, alert thresholds and integrations to the needs of a specific critical infrastructure (CII) use case. The central proposition is simple: a common platform can provide consistency, but its detection and response logic must be configurable to reflect the environment it protects. This is an illustrative design case study, not a claim about a named customer deployment or measured product performance. The scenarios and expected outcomes below are analytical examples intended to show how customisation can improve relevance when implemented, validated and governed appropriately.
“A common security platform may scale across sectors. Effective OT protection depends on whether it can be configured to understand what is normal, abnormal and consequential for each asset.”
1. The Standardisation Paradox
Organisations often seek standardisation to simplify deployment, centralise monitoring and compare risk across sites. Those are legitimate objectives. The challenge appears when standardisation becomes rigidity: the tool ships with fixed assumptions about assets, protocols, expected traffic, severity and alert logic, and local teams have limited ability to adapt those assumptions. What standardisation helps with Where a fixed model can fall short
May not capture the operational role or consequence of a Consistent inventory and telemetry collection particular asset May rank technically similar events identically despite different Common alert format and reporting process impacts May mistake legitimate maintenance or engineering activity for Reusable baseline detections malicious behaviour—or miss a sector-specific deviation Centralised dashboards May hide local context behind generic severity scores May not accommodate legacy devices, unusual protocols, site- Repeatable deployment specific topology or strict change windows
The goal is not to abandon standardisation. It is to standardise the platform’s foundations while allowing controlled customisation of the detection and operational context layers.
Figure 1. Standardised foundation, configurable context layer: the platform core stays common while sector and site context
is configured on top.
2. Critical Infrastructure Is Not One Threat Environment
Sector and site context changes what defenders should prioritise. The following examples are illustrative; the actual technology and threat profile must be validated for each facility. Potentially distinctive assets and Examples of context-specific CII sector / use case dependencies security questions
Is a controller communicating with an Baggage handling PLCs, building unexpected engineering station? Could Airport operations management, airfield lighting, passenger- an alert affect baggage throughput or processing dependencies airfield operations? Is a control command expected for this Protection relays, RTUs, substation Electricity / utilities operating state? Is the source authorised automation, SCADA links for this substation and protocol? Does a setpoint change fit the approved Chemical dosing, pumps, pH Water treatment operating envelope and current work instrumentation, treatment SCADA order? Could the event affect process safety, DCS, SIS, compressors, valves, process Oil, gas and process industry production continuity or a safety historians instrumented function? Signalling interfaces, traction power, Is the communication permitted for this Rail and transport station systems, control centres signalling zone and operational mode? Is this firmware, logic download or Robots, CNC equipment, PLC lines, Manufacturing machine-state transition expected for this industrial gateways production cell?
A sector label alone is not enough. Two airports—or two plants in the same sector—can differ in architecture, vendor mix, operating schedules, remote-access arrangements, redundancy and tolerance for disruption. Effective configuration must therefore work at both sector and site level.
3. The Case for a Customisable OT Security Tool
Customisation should be more than changing dashboard colours or setting a generic alert threshold. It should allow a security team to shape how the platform models the environment, recognises relevant behaviours, interprets risk and delivers evidence to the people responsible for the process. Customisation capability Why it matters Example
Connects device identity to function, Distinguish a baggage-line PLC from a Asset and process model zone, owner, dependencies and non-critical display controller consequence Supports diverse OT stacks, variants and Decode the relevant S7, Modbus, OPC UA Protocol and vendor support vendor-specific behaviours or BACnet traffic where present Allows rules to reflect local normal Alert on a PLC logic download outside an Detection logic and rules behaviour and threat scenarios approved maintenance window Treat a scheduled vendor session Accounts for legitimate cycles, shifts, Site-specific baselining differently from an unknown session at seasonal demand and maintenance an unusual time Combines technical signal with Escalate a low-volume event on a safety- Risk weighting reachability, criticality and process critical asset above a noisier event on a impact redundant monitor Maps sector-relevant adversary tactics to Prioritise remote-access and engineering- Threat intelligence mapping assets and exposed pathways workstation pathways relevant to the site Place a clearly governed decoy Enables decoys and breadcrumbs to Deception and custom use cases representing a likely engineering service reflect the protected environment or OT asset class Routes actionable context to SOC, Send enriched alerts to SIEM/SOAR with Workflow and integrations engineering, operations and incident asset role, evidence and recommended response validation
Figure 2. Reference architecture for a customisable OT security platform, from passive collection to enriched alerts, under
configuration governance.
The platform should make these capabilities configurable without requiring every customer to rebuild the product or write unreviewed code. Configuration should be versioned, testable, reversible and subject to change control.
4. Illustrative Scenario: One Platform, Three Different Contexts
Consider the same technical signal: a controller receives a write command from an engineering workstation. A generic rule may flag the command based on protocol function alone. A context-aware and customisable tool can interpret the event against approved sources, asset role, time window, operating state and process consequence. Environment Context supplied to the tool More useful detection outcome
Prioritise an otherwise valid write PLC identity, line dependency, approved command when it originates from an Airport baggage handling vendor endpoints, maintenance schedule, unapproved host or outside the approved impact on baggage flow window Controller role, approved setpoint range, Correlate an unusual setpoint write with Water treatment process state, dosing safeguards, work process limits and source identity; alert at order the appropriate urgency Machine recipe, production shift, Distinguish an approved recipe change Manufacturing cell authorised engineering station, change from an unexplained logic download or ticket out-of-pattern write
Figure 3. One technical signal, three operational meanings: context changes how the same write command is interpreted
and prioritised.
The signal is technically similar; the operational meaning and response are not. Configuration allows the platform to preserve a common detection foundation while tailoring its interpretation and prioritisation to each CII environment. This does not mean context should suppress alerts automatically. Where evidence is incomplete, the tool should expose uncertainty, retain the raw event and allow escalation rather than silently treating an unfamiliar event as benign.
5. From Generic Alerting to Use-Case-Led Detection
A practical implementation begins with defined questions—not with an unrestricted list of features. Each use case should specify the asset or process in scope, the behaviour of interest, the relevant telemetry, the expected baseline, the threat hypothesis, the consequence and the response owner. Use-case element Questions to define before enabling a detection
Operational objective What process, service or safety function are we protecting? Which devices, zones, communication paths and dependencies Asset scope are in scope? What plausible misuse, intrusion path or unsafe change are we Threat hypothesis trying to detect? Which packets, logs, identity events, configuration changes or Required evidence process signals support the detection? What legitimate behaviour is expected, and who approves Baseline and exceptions exceptions? Who must act, within what time, and what action is safe for Severity and response operations? How will the rule be tested against representative benign and Validation malicious scenarios? Who owns the rule, reviews drift, approves updates and can roll Governance back a change?
This use-case-led approach reduces the risk of deploying broad rules that generate large volumes of low-value alerts while missing behaviours unique to the environment.
6. Customisation Must Be Governed
Flexibility is valuable only when it remains controlled. Excessive or poorly documented customisation can fragment deployments, weaken detection consistency and create rules that nobody understands or maintains.
Figure 4. Governed configuration lifecycle: propose, test, stage, approve, deploy and review, with rollback and periodic
reassessment.
• Use approved templates and a common core of baseline detections, then add sector- and site-specific rules. • Version every configuration and record the owner, purpose, evidence source, approval and change history. • Test rule changes against representative benign traffic, known attack scenarios and historical events where available. • Use staged deployment or monitor-only mode before enabling high-impact alerting or automated response. • Define expiry and review dates for exceptions; avoid permanent allowlists without justification. • Keep raw telemetry and an auditable explanation of why an event was classified, prioritised or suppressed. • Ensure response recommendations are OT-safe. Do not automatically isolate or alter control devices without an approved operational safety process. • Reassess configurations when assets, network paths, vendors, production processes or sector threat conditions change.
7. How to Measure Whether Customisation Works
The value of customisation should be demonstrated through validation, not assumed from the number of available settings. Baseline and customised configurations should be compared using the same defined test scenarios and comparable data. Measure What it indicates Evaluation approach
How many alerts correspond to validated Review alert outcomes against labelled Detection precision events of interest test cases or analyst adjudication How often benign activity triggers the Run approved maintenance, normal False-positive rate detection production and routine traffic scenarios Which defined threat scenarios are Map tested cases to the use-case register Detection coverage detected and which remain uncovered and document gaps How quickly an event is identified and Measure event-to-alert and alert-to-triage Time to detect / triage understood intervals under comparable conditions Whether alerts include asset role, zone, Audit a sample of alerts against required Context completeness dependencies and operational context fields consequence Whether configurations can be Review ownership, documentation, Rule maintainability understood, tested and safely updated version history and rollback readiness Validate passive monitoring and response Whether monitoring and response avoid Operational safety actions with OT engineering and safety unacceptable process disruption teams
No numerical improvement is claimed in this illustrative case study. Organisations should establish their own baseline, define acceptance criteria and publish results only after repeatable testing or measured deployment.
8. A Practical Adoption Roadmap
Phase Activity Deliverable
Inventory assets, protocols, zones,
1. Discover Validated asset and process context
owners and critical dependencies Choose a small number of high-value,
2. Select Prioritised use-case register
sector-relevant use cases Adapt parsers, baselines, rules, context
3. Configure Versioned configuration pack
fields and workflows Test benign, suspicious and approved Evidence-backed test report and known
4. Validate adversarial scenarios in a safe
limitations environment Integrate alerts with SOC, OT
5. Operationalise Defined triage and escalation workflow
engineering and incident response Review outcomes, changes in assets and Controlled tuning and periodic
6. Improve
evolving threat scenarios revalidation
Figure 5. Adoption roadmap: six phases from discovery to continuous improvement, each with a defined deliverable.
A small number of well-validated use cases is usually a more defensible starting point than enabling every available rule. Expansion should follow evidence of value, operational readiness and maintainability.
9. What Buyers Should Ask Their OT Security Provider
• Can we model our own asset roles, process dependencies, zones and consequence levels? • Can detections be configured for our sector, site, vendor mix and operational schedule? • Which protocol variants and vendor-specific behaviours are actually supported, and how is that support validated? • Can our team build, test, version, approve and roll back a custom use case without a product-code change? • Can alerts explain the evidence and context behind their priority instead of exposing only a score? • How are false positives, exceptions and changes in normal operating behaviour managed? • Can the platform integrate with our SOC while preserving OT engineering authority over operational actions? • What evidence can the provider share for detection coverage, false-positive performance and safe deployment?
Conclusion
Critical infrastructure is diverse by design. The same protocol can serve different processes; the same alert can carry different consequences; and the same sector can contain sites with materially different architectures and threat exposures. A one-size-fits-all OT security model risks flattening those differences into generic detections and scores. The stronger approach is a consistent platform with meaningful, governed customisation: sector-aware threat scenarios, site-specific asset and process context, adaptable detection logic, explainable prioritisation and workflows aligned with the people who operate the infrastructure. For CII operators, the key question is therefore not simply whether an OT security tool can see devices or recognise protocols. It is whether the tool can be configured—and safely maintained—to understand the specific environment, detect the behaviours that matter there, and present evidence that supports timely operational decisions.
A position, not a product claim
This case study presents a general design perspective on customisable OT security. The scenarios are illustrative, and the benefits described are expected outcomes to validate rather than measured results from a named deployment. Actual effectiveness depends on data quality, architecture, configuration, testing, governance and operational adoption.
