Heavy-Industry Edge Data Fabric: Policy-Aware Governance and Industry Integration into EOSC

Heavy‐industry sites - ranging from quarries to ports - are increasingly deploying multimodal sensors to capture environmental, operational and logistics data. However, these rich data streams often remain siloed, hampering both operational optimisation and environmental stewardship. We propose a Heavy-Industry Edge Data Fabric that embeds policy-aware data governance, hybrid compute and FAIR principles directly at the source, enabling heavy-industry facilities to become active nodes in the European Open Science Cloud (EOSC) ecosystem.

Traditional heavy-industry networks often struggle with latency and bandwidth constraints when transferring high-volume, time-critical data to centralised analytics centres. Furthermore, commercial confidentiality and safety regulations frequently mandate that raw data remain on-site, and there are no mechanisms to distinguish between sensitive and sanitised datasets. Our framework addresses these challenges by deploying Edge-to-Fog aggregation nodes at such facilities. These nodes comprise ruggedised IoT sensors – such as monitoring dust, noise, vibration, weather, vessel movements and logistics feeds – with data fed into a local Fog-layer governance engine built on iRODS. This engine applies Attribute:Value:Unit (AVU) tagging and enforces granular policy rules to segregate open (publicly shareable) from sensitive (proprietary or regulated) data at ingestion.

A centralised orchestration layer aligned with HEAnet’s eduGAIN AAI, integrates with facility-level gateways and EOSC-Ireland services to authenticate users and manage open-data flows. Through containerised Composable Open Data & Analysis Services (CODAS), heavy-industry operators can request end-to-end data pipelines, such as scheduling transfers of processed environmental metrics for open science initiatives shared on EOSC repositories. Additionally, sites can participate in open science and environmental monitoring by launching hybrid compute jobs on EOSC-linked Clusters, in collaboration with university research partners. Each CODAS component - Edge ingest, Fog policy engine, DCAT-AP exporter, workbenches, discovery interface, and agentic AI assistants - is packaged with Helm charts, continuous-integration tests and software bills of materials (SBOMs), ensuring reproducibility and portability across quarries, ports, mines or manufacturing plants.

 This approach would not only accelerate sustainable practices and regulatory compliance across heavy-industry sectors but would also enrich European research infrastructures with unique multimodal datasets - bridging the gap between private operators and open-science objectives.