From paper to product, in five steps.
One concrete case, end to end: the B737-800 Environmental Control System sensor reduction. The problem, the method (MOSOF + NDCI), the real Pareto data the algorithm produced, the venture it turned into (AcoustR), and the peer-reviewed paper that came out.
An ECS with too many sensors and not enough useful ones.
The Boeing 737-800 Environmental Control System — the airframe's air-conditioning and pressurisation suite — is a tightly-coupled assembly of compressors, heat exchangers, valves, and turbines. Its sensor network grew the way most aerospace networks grow: incrementally, each subsystem owner adding the sensors they need, no one owning the whole picture.
The consequence is the consequence everyone knows: a network simultaneously over-instrumented in some places (redundant sensors that contribute nothing to a diagnosis) and blind in others (failure modes nothing currently sees).
Status quo
Reference sensor suite designed by accumulation, not by optimisation. No measure of which sensor is "earning its place".
≈ 18 sensorsThe question
Could a smaller, lighter, cheaper sensor suite match — or exceed — the diagnostic capability of the larger one?
12 / 0.69MOSOF + NDCI: a framework, then a score.
The Multi-Objective Sensor Optimisation Framework (MOSOF) treats sensor selection as a constrained search over the asset's information surface. Instead of asking "is this sensor good?" — a question with no answer — it asks "given everything we care about (cost, weight, reliability, diagnostic coverage), which combinations of sensors are non-dominated?" The output is the Pareto front, and a stakeholder picks from it explicitly.
The Normalised Diagnostic Contribution Index (NDCI) — the second contribution — gives each candidate sensor a comparable, normalised score for its share of system-level diagnostic capability. Two sensors that look similar on paper can contribute very differently to actual fault diagnosis, and the difference is invisible without a measure like NDCI. NDCI fixes the long-standing ambiguity of "more sensors = better".
The Pareto front the algorithm actually produced.
Three objectives — diagnostic performance, cost, suite reliability. Every point shown is non-dominated. The recommended knee is the published 12-sensor suite (Engine 5 / Fuel 2 / EPS 2 / ECS 3) at 0.69 normalised diagnostic performance, $36k, 145 kh suite MTBF. Drag the sliders to re-weight axes — the recommendation moves accordingly.
AcoustR: the same thinking, taken to its limit.
AcoustR — my lead venture — answers the diagnostic-value-per-sensor question by going to the limit: a single, low-cost, high-information sensor (a microphone). Same NDCI thinking, applied to acoustic monitoring of engines and machines. The product is live; the link below is the easiest way to see it running on real data.
Acoustic diagnostics for engines and machines
Microphone in, condition assessment out. The diagnostic-contribution thinking from MOSOF + NDCI, applied to the hardest sensor of all: the one you can put on anything that vibrates.
See it liveAnd the peer-reviewed paper that came out.
The thesis itself — Sensor Optimisation for Aircraft Health Management Systems, Cranfield University 2025 — is the canonical document for the cross-subsystem evaluation, including the validation chapters that produced the Pareto front above. Embedded inline below; downloadable in full from the link if your browser blocks PDF embedding.
The associated journal article — Suslu, Ali, Jennions (2026). MOSOF with NDCI: A Cross-Subsystem Evaluation of an Aircraft for an Airline Case Scenario. Sensors 26(1), 160. doi:10.3390/s26010160 ↗
If your problem looks like this one, I'd like to hear about it.
Multi-objective decisions, sensor networks that grew rather than were designed, hard trade-offs between performance and cost — those are the problems this method is built for.
[email protected] →What this proof enables.
Peer-reviewed methods, published code, and validated case studies are not just academic credentials. They are the reason a consulting engagement or a technical hire can be scoped, evaluated, and trusted without a leap of faith.