Home / Resources / What is NDCI?

What is NDCI?

A plain-English explainer for the Normalised Diagnostic Contribution Index. The technical paper is in Sensors; this page is for the reader who wants to understand what NDCI does, why it's useful, and what it changes about how sensor networks get designed.

TL;DR

NDCI is a per-sensor score for the share of system-level diagnostic capability that sensor contributes. It lets candidate sensors be compared on a common, normalised scale instead of by domain intuition. The practical effect: instead of arguing about whether a sensor is "good", you can rank candidates by how much they actually move the diagnostic-coverage needle on the system you care about.

Commercial translation

NDCI gives engineering teams a defensible, auditable score for every sensor in a network — based on how much it actually contributes to diagnosing faults, not on historical installation decisions. This turns "which sensors should we keep?" from a committee argument into a ranked, reproducible analysis.

Typical application: mid-life sensor review, pre-tender configuration justification, IVHM programme scoping, or any situation where the cost of instrumentation must be justified against its diagnostic value.

The acronym

NDCI stands for Normalised Diagnostic Contribution Index. It was introduced in Suslu, Ali, Jennions (Sensors, MDPI, 2025) as a per-sensor score that integrates with the Multi-Objective Sensor Optimisation Framework (MOSOF). The 2025 paper validated the combined MOSOF + NDCI approach on the Boeing 737-800 Environmental Control System through Cranfield's SESAC simulator.

What problem it solves

"Is this sensor good?" is a question with no defensible answer in isolation. A sensor that is excellent for one failure mode may be useless for another, and two sensors that look identical on paper can contribute very differently to a real system-level diagnosis.

Before NDCI, sensor-design decisions usually came down to a mix of "more is better" intuition, dominance arguments (this sensor is strictly better than that one), and engineering experience. None of those gives a comparable, normalised score across candidates. NDCI does — and that turns sensor selection from a debate into a ranking.

Before NDCI

"Sensor A and Sensor B both detect this failure mode — let's add both." Result: redundancy, weight, cost, no comparable measure of who actually helps the diagnosis.

With NDCI

Each sensor gets a normalised score for diagnostic contribution. Sensor A might score 0.42; Sensor B 0.08. The decision becomes informed instead of intuitive.

How it works (high level)

  • Define the diagnostic surface. What failure modes do we care about, and what's the system-level diagnostic capability we're measuring against?
  • Score each sensor's contribution. For every candidate sensor, NDCI quantifies how much that sensor contributes to the system's overall diagnostic capability — not in isolation, but conditional on the rest of the network.
  • Normalise. Scores are normalised so that sensors of different physical types and performance characteristics can be compared on a common scale.
  • Feed into MOSOF. NDCI's per-sensor scores plug directly into MOSOF's multi-objective search — one of the objectives the framework optimises over is total network NDCI.
Fig. 1 NDCI contribution bars across seven candidate sensors. Toggle a fault mode off to see how the ranking shifts — sensors that scored high on one fault drop when that fault leaves the diagnostic problem. Illustrative scores for a stylised rotating-machine problem, not from any single MOSOF run.
Fig. 2 The same data as a heatmap. Cells with the rust ring are best-in-column — the highest-scoring sensor for that fault. The four best-in-column sensors here are not unique: oil debris and pressure dominate one fault each, while vibration and speed share two. Illustrative; matrix as in Fig. 1.

What it changes about sensor design

The biggest shift NDCI enables is a reliable answer to the question of diminishing returns. With a per-sensor diagnostic-contribution score, you can see when adding the next sensor stops being worth the cost, weight, and reliability burden. That answer used to be a judgement call — NDCI makes it a calculation.

The validated result on the Boeing 737-800 ECS: smaller, lighter, cheaper sensor configurations that match or exceed the diagnostic capability of larger reference designs. That's not theoretically nice — it's the kind of answer an OEM, operator, or MRO would actually want to install.

Want NDCI on your network?

The Sensor Optimisation Audit applies NDCI scoring to an existing sensor network in three weeks. The full consulting page has details on what's included and what comes back.

See the consulting page →
Related termWhat is MOSOF? →
Applied ventureAcoustR →
Direct enquiryEmail Burak