An open source digital manufacturing platform can help manufacturers connect production systems, industrial data and business applications without treating every factory technology as a separate island. The MIDIH reference architecture sets out a flexible, modular approach for Industry 4.0, combining edge-oriented factory automation, interoperability for brownfield environments and industrial analytics. Its design offers useful guidance for engineers and decision-makers planning how data should move from machines to local systems, enterprise services and supply chain partners.
The aim is not simply to collect more data. A manufacturing platform must support the movement, management and interpretation of information across different operational levels while providing interfaces to existing and third-party systems. MIDIH organizes these concerns into an architecture that can support smart factory, smart product and smart supply chain implementations.
Why a shared industrial reference architecture matters
Industrial IoT projects often need to bring together technologies and systems developed under different frameworks. MIDIH draws on four reference architectures: FIWARE, the Industrial Data Space (IDS), the Industrial Internet Reference Architecture (IIRA) and Plattform Industrie 4.0’s RAMI 4.0. The project’s goal is to align concepts from these models in a meta-model that can support interoperable reference implementations.
For industrial teams, this kind of alignment helps frame architecture decisions across connected systems. It gives a project a way to consider data, functions and interactions across factory, product and supply chain use cases, rather than designing each connection in isolation. It does not remove the need to assess individual interfaces or deployment requirements, but it establishes a common architectural context for doing so.
Open source digital manufacturing platform architecture
MIDIH extends the open source BEinCPPS platform with additional components and capabilities. These include distributed, edge-oriented factory automation; support for brownfield interoperability; and an industrial analytics platform built from open source components for cyber-physical systems and IoT data management.
The architecture follows a data-driven approach inspired by the IIRA Layered Databus pattern. It distinguishes between levels where data is produced and used:
- Unit and edge level: Sensors, devices, actuators, machines and products generate operational data. At this level, a databus may be embedded within a smart system such as a cyber-physical production system or robot.
- Site level: Data and interactions are managed within a factory or organizational site. Depending on the deployment, the site and unit levels may overlap; the unit level can also represent edge-cloud layers associated with a production line or assets within a department.
- Inter-site level: Secure data exchange supports interactions across factories, enterprises and value chains.
This separation helps teams think about where processing belongs and how information should travel. MIDIH describes edge computing through local clouds and cloud computing through hybrid clouds, with data-in-motion and data-at-rest treated as distinct architectural concerns.
Connecting operational data with analytics
At the platform level, the architecture includes data access and information services such as APIs, microservices and software development kits. The data-in-motion path includes an event broker, visualization and services, alongside security and field gateway functions. These capabilities are associated with industrial IoT and edge computing.
Data-at-rest capabilities include an information bus, semantic models, persistence, visualization and services. The design also identifies offline processing and analytics middleware. Together, these elements cover ingestion, integration, storage and analysis, while separating the handling of streaming information from stored data.
That distinction matters for manufacturing applications. Processing data in motion supports real-time analytics and event detection using streaming information and edge computing techniques. Processing data at rest supports analysis of persisted data, metadata and models. The platform’s described analytics services include exploratory and multivariate analysis, predictive analytics and deep learning, while visualization helps users interpret results for decision-making.
Interoperability with existing systems
A manufacturing architecture rarely operates on open source components alone. MIDIH identifies open APIs as a way to connect its platform with proprietary cloud manufacturing, product lifecycle management, enterprise resource planning and IoT solutions. The examples named in the project material include Siemens MindSphere, Dassault Systèmes 3DEXPERIENCE, SAP HANA and GE Predix.
For system integrators, the practical point is that interface design is part of the architecture, not an afterthought. APIs and shared data models can help connect systems with different roles, while brownfield interoperability addresses the reality that factories may contain established equipment and technologies that cannot simply be replaced. The specific integration work will depend on the systems and interfaces involved.
Choosing components and implementation lanes
MIDIH’s comparison of implementation lanes highlights shared objectives: flexibility, agility, the ability to start small, shared maintenance costs, interoperability, scalability and cost-effectiveness. It also contrasts FIWARE and Apache-based approaches. FIWARE is described as providing standard APIs for context information and a common layer for exchanging contextualized data across varied IoT protocols. Some FIWARE generic enablers are based on Apache components.
The material also notes trade-offs. It identifies limited vendor expertise, platform security and comprehensiveness, and fewer applications in real environments as concerns for FIWARE. Apache, meanwhile, is described as having a stronger community, which can contribute to component support and evolution. These considerations are prompts for evaluation, not a universal ranking: teams should assess their own expertise, integration needs and governance requirements when selecting components.
From architecture to industrial use cases
The MIDIH architecture is intended to support connected capabilities across three domains:
- Smart factories: Data ingestion and transfer, streaming analytics, persistence, data analytics and visualization.
- Smart products: Data-driven services and analysis that can help users contextualize results and make decisions.
- Smart supply chains: Standardized collaboration between OEMs and subcontractors, visibility into production, inventory and materials, and analytics-supported planning.
For supply chain data sharing, the architecture identifies mechanisms based on digital identity, sharing policies, sharing agreements and data certification. These elements reflect the need to manage not only connectivity but also the conditions under which information is shared.
Practical implications for engineering teams
MIDIH’s central lesson is that a digital manufacturing platform needs to be designed as a connected system of data flows, services and interfaces. Edge and cloud processing, data-in-motion and data-at-rest management, industrial analytics and secure inter-site exchange are related architectural concerns. Addressing them together can give implementation teams a clearer basis for connecting production environments with enterprise and supply chain systems.
Before deployment, teams still need to map their existing assets, identify required integrations and choose components that fit their operational context. A modular, open source reference architecture offers a structured starting point: it can support incremental implementation while keeping interoperability and future connections in view.
Access Official Technical PDF