Distributed Fibre Sensing
Continuous observation across distance using the ultra-light fibre infrastructure.
The Proposition
IDRCIN is not proposed to add more technology to Imbak Canyon. It uses a Fibre-First Architecture, adding selective point instruments only when a research requirement cannot be adequately met through fibre or remote observation.
Research Infrastructure
Designed for a living rainforest
Imbak Canyon is treated as a conservation and research environment first. Technology remains subordinate to scientific and ecological priorities.
Operating Principle
Human when necessary
Field science remains essential where physical sampling, ecological judgement or ground-truthing is required. IDRCIN reduces unnecessary human presence; it does not replace researchers.
“Bring the instrument to the forest without routinely bringing people beneath the canopy.”
System Architecture
The architecture keeps the fibre corridor primary, ultra-light and relocatable. An Autonomous Specialised Point-Sensing Layer extends research capability only where a physical probe is scientifically justified.
Layer 1 · Fibre-First ResearchDrone survey → ultra-light fibre corridor → Distributed Fibre Sensing through DAS, DTS, DSS and applicable backbone functions → retrieval and relocation.
Layer 2 · Selective Point SensingSparse Autonomous LoRa Scientific Pods measure only research-defined properties that fibre cannot adequately provide, with local buffering and low-duty-cycle transmission.
Layer 3 · Research IntelligenceObservatory / HQ → deterministic QA/QC → Temporal Digital Twin → NeuralOps when useful → validated analytics and authorised research access.
One Observatory, Two Complementary Sensing Modes
Fibre remains primaryAutonomous Specialised Point Sensing
Sparse battery-powered scientific pods, one specialised sensor per pod, used only where a physical probe is required.
Fibre senses continuously across space. Autonomous pods measure specific properties at selected points.
Scientific SensorResearch-specific probe
Low-Power ControlDeep sleep and sampling
Local BufferConnectivity loss is not data loss
Low-Duty LoRaEvent or summary payload
Research PlatformHealth, retrieval and redeployment
Zero New Terrestrial Gateway Infrastructure: no new gateway tower, repeater, mesh, communications hut or permanent intermediate wireless station inside the protected forest. An existing or approved receiver may be used where technically feasible and where link budget permits. Satellite / NTN remains a candidate subject to canopy, terrain, link-budget, regulatory and field validation.
Recon Drone
3D Digital Twin
AI Route
Drone Deploy
Research Observatory
TM Cloud / HQ
Research Intelligence
NeuralOps Detached Systems validate distributed fibre streams and received or recovered pod records. Deterministic checks run first; AI reasoning is used only where it adds research value.
Raw Data
Original measurements are preserved, including locally buffered pod records received later by radio or retrieval.
Validated Data
Noise, duplicates, timestamp issues, drift and suspicious values are flagged through auditable rules.
Projection
Historical, received and delayed patterns can generate advisory projections with confidence, assumptions and time horizon shown.
Raw data is evidence. Cleaned data is operational. Indicators are interpreted information. Projections are advisory.
Presence vs Impact
The system does not claim zero impact. It asks a harder question: which method produces the required scientific value with the lowest reasonable total ecological disturbance?
Low
Routine human presence
Routine deployment, inspection and retrieval are designed for drone operation, while human fieldwork remains available whenever science or safety requires it.
High
Spatial flexibility
Sensor sets can rotate between research zones after six or twelve months, expanding cumulative coverage without permanent instrumentation at every site.
Can fibre provide it?Yes: use Distributed Fibre Sensing. Do not add a pod merely because wireless sensing is possible.
If no, test the research needEvaluate scientific value, payload, battery, calibration, ecological presence and reliable retrieval.
Do-Not-Deploy PrincipleIf scientific value is insufficient or the instrument cannot be responsibly managed, do not deploy.
Wireless capability does not create permission to deploy. Every Autonomous LoRa Scientific Pod must justify its ecological presence through a defined scientific requirement.
ESG & Carbon
IDRCIN accounts for presence, energy and carbon honestly measured, not assumed.
IDRCIN vs Manual
Less routine presence
Drone-deployed, Retrievable / Relocatable Instrumentation can reduce repeated human access, while pod batteries, electronics and retrieval missions remain part of the measured footprint.
NeuralOps vs Full AI
Local, on-demand intelligence
Validation runs on-premise via NeuralOps Detached Systems; heavy cloud LLM is used sparingly, keeping energy and carbon proportional to need.
Carbon footprint is budgeted and disclosed including grid electricity (e.g. TNB) for research systems, measured rather than assumed away.
Two Layers of Operational Carbon Reduction
IDRCIN targets carbon reduction at the physical research layer by reducing repeated field mobilisation. NeuralOps targets carbon reduction at the digital intelligence layer by reducing unnecessary AI processing.
“Reduce unnecessary movement in the forest. Reduce unnecessary computation in AI.”
Physical Research Layer
IDRCIN vs Conventional / Manual Monitoring
Preliminary scenario estimate
Manual Monitoring
12 campaigns2 × 4×4
7,200 km × 0.256 = 1,843.2 kg
0
tonnes CO₂e / year
VS
IDRCIN
6 inspections1 × 4×4
drone charging · 180 kWh / yr
460.8 + 97.0 = 557.8 kg
0
tonnes CO₂e / year
≈ 70% LOWERoperational field emissions
This comparison focuses on operational field emissions, primarily ground transport and drone electricity. It does not yet include full embodied-carbon lifecycle emissions from manufacturing vehicles, drones, sensors, fibre, batteries or infrastructure. Actual project values should later be replaced with measured vehicle kilometres, fuel litres, drone battery charging kWh, field mission count and retrieval missions.
Digital Intelligence Layer
NeuralOps vs Full-AI Processing
Working workload comparison
Full AI
large AI workload32B tokens
VS
NeuralOps
after routing / detached filtering2.5B tokens
≈ 92.2% LESSvariable AI workload
Whole-system energy model: 30% fixed infrastructure + 70% variable. NeuralOps = 30% + (70% × 2.5/32) = 35.47%.
≈ 64.5% LOWERestimated whole-system compute footprint
Illustrative example based on an existing 360 kg CO₂e annual Full-AI baseline: Full AI ≈ 360 kg; NeuralOps ≈ 128 kg; estimated avoided ≈ 232 kg CO₂e / year. Not externally audited data.
IDRCIN reduces repeated physical mobilisation.
NeuralOps reduces unnecessary AI computation.
NeuralOps reduces unnecessary AI computation.
More scientific intelligence with less operational overhead efficiency at both the forest edge and the compute layer.
From Estimated → Measured
Physical Layer
- vehicle kilometres
- fuel consumption
- drone battery kWh
- number of missions
- human field hours
Digital Layer
- total tokens
- model calls
- server electricity
- cloud workload · storage · network
Future KPI
kg CO₂e / research pointkg CO₂e / month of monitoringkg CO₂e / GB validated datakg CO₂e / research output
Goal: replace preliminary scenario estimates with measured operational ESG data during POC / pilot.
Balanced SWOT
Every advantage is paired with its limitation and a response strategy. Robustness comes from layered design, not claims of perfection.
Strengths
- Drone-first, low routine human entry
- Fibre-first distributed sensing with selective point measurements
- Pod-local buffering during communication interruption
- Retrievable and reusable research layer
- Zonal intelligence and layered storage
- Temporal Digital Twin and projections
- IDRCIN Research Observatory
Weaknesses
- High R&D integration complexity
- Canopy mapping remains imperfect
- Fibre, pod enclosure and retrieval behaviour require field proof
- Battery lifecycle, calibration and RF uncertainty
- Drone endurance constraints
- Projection accuracy requires historical validation
Opportunities
- Rainforest microclimate and biodiversity research
- Hydrology and climate resilience studies
- Research-as-a-platform for multiple institutions
- Long-term Sabah environmental intelligence
- Replication to other conservation landscapes
Threats
- Extreme weather and wildlife interaction
- Regulatory limitations
- Canopy attenuation, terrain shadowing and cyber risk
- Technology obsolescence
- Scaling beyond ecological justification
Explore 60 research programmes in the IDRCIN Research Observatory.
Key Risks & Response
Critical risks are designed into the operating model rather than hidden from the proposal.
Fibre snaggingBranch movement, abrasion or entanglement.Canopy-adaptive routing, controlled slack, retrievability scoring and tension-controlled recovery.
Bird / wildlife interactionCollision, curiosity, pulling or biting.High-visibility fibre candidates, field observation and ecological validation before scale.
Automated cleaning errorValid extreme data may be misclassified.Raw data preserved; suspicious values flagged rather than silently deleted.
Projection errorForecast may be wrong.Projection remains advisory, with confidence, assumptions and supporting evidence visible.
Pod power / enclosureBattery depletion, moisture ingress, corrosion or sensor drift.Low-duty operation, battery-health telemetry, environmental testing, calibration and planned retrieval.
Connectivity failureCanopy attenuation, terrain shadowing, receiver, Internet or cloud interruption.Pod-local buffering first, then observatory storage and synchronisation after connectivity or physical retrieval.
Recommended Pilot
IMBAK recommends a small Phase-2 Autonomous LoRa Scientific Pod pilot before any scale decision. Pods remain a separate scientific instrument pool until asset-level reconciliation is completed.
Joint Workshop
Recon Baseline
Route Review
Small Pod Pilot
Observe & Validate
Retrieve & Review
Engineering KPI
RF feasibility, battery consumption, enclosure survivability, local data retention and retrieval performance.
Scientific KPI
Sensor accuracy, calibration stability, data completeness, traceability and unique value beyond fibre.
Ecological KPI
Physical footprint, wildlife interaction, maintenance interventions and complete post-pilot recovery.
Pilot decision gate: GO / MODIFY / STOP. Allocation or reallocation must remain within the locked RM2.0M Phase-2 budget envelope after BOM and field validation; no LoRa hardware price is assumed here.
More knowledge per unit of ecological presence.
IDRCIN is designed to help Yayasan Sabah and research partners understand Imbak Canyon more continuously, more spatially and more intelligently while reducing unnecessary physical intervention wherever practical.
Imbak Canyon Conservation Area · Sabah
Continuous, spatially distributed scientific knowledge with minimal physical presence beneath the canopy.
