FabOS: An Open, Distributed, Real-Time Capable, and Secure Operating System for Production

Industry 4.0 promises to create the fourth industrial revolution and digitally transform production into smart manufacturing. New and ever-improving technologies are expected to improve flexibility and changeability for manufacturing systems. However, rigid, functionally separate IT architectures and fear of making data available for artificial intelligence (AI) applications still limit flexibility and use of data-driven technologies.

The FabOS research project aims to resolve these issues with an AI-focused operating system for production that is open, distributed, real-time capable, and secure. The project is funded by the German Federal Ministry of Economic Affairs and Energy (BMWi) as part of the 2019 AI innovation competition and combines the efforts of 22 partners, including the Eclipse Foundation, research institutions, universities, and companies.

FabOS Enables Cyber-Physical Production

The FabOS operating system should not be thought of like a PC operating system or comparable dedicated operating systems, but as a system of orchestrated components and services for the operation of a networked factory that consists of cyber-physical production systems. Together, these systems shape the factory as a cyber-physical production operation.

The production operating system forms the IT backbone for versatile automation in the factory of the future and the basis of an ecosystem for data-driven services and AI applications. The base framework of the operating system is the meta-kernel that provides the core functionality, similar to the kernel of an operating system.

FabOS integrates technologies that improve the real-time proximity of existing applications and enable and guarantee the hard-real-time capability and determinism of real-time systems in a modern and flexible infrastructure.

Open Source and Open Standards Are Key Enablers for Open Systems

Today, there’s a trend to virtualize hardware functions, which means software now lives longer than hardware. Companies are also increasingly using many innovative software products that are developed by open source communities. Much of the progress that has been made in data science and AI research and in application development over the last few years was enabled by open source tools that were provided to a broad community. As a result, open source has become one of the most important innovation drivers.

The FabOS project aims to follow the established principles of open source innovation, which includes an open architecture, integration of accepted standards, and strong stakeholder involvement. The project partners are developing an open architecture for FabOS and are implementing it as open source.

The FabOS architecture is based on existing and established reference architectures from various domains and integrates common and open established standards. Open standards such as HTTP, SMTP, MIME, and OpenDocument have accelerated the success of the internet and its associated web technologies. Similarly, web technologies have been successfully applied to develop distributed systems based on a service-oriented architecture (SOA) and are widely accepted in many industrial applications.

Data Governance for Data-Driven Technologies

FabOS integrates multiple standards and technologies and is adopting upcoming standards, such as the Asset Administration Shell (AAS), which is expected to become the unifying standard for interoperability in Industry 4.0. By adopting a unifying standard to virtually represent every asset in the digital production environment, the data governance that is needed for efficient use of data-driven technologies can be put in place.

Figure 1 illustrates the key criteria for data governance.

Figure 1: Key Criteria for Data Governance in Digital Production Environments

(No Description)

Because FabOS is based on an open architecture with open interfaces, the components integrated into it can be open source or they can be proprietary technology. This flexibility underlines the importance of adhering to open standards and interfaces to enable the simple exchange of components based on their compatible open interfaces.

FabOS adopters can select components from a broad range of open source and proprietary solutions, and even develop their own components where needed. As a result, the FabOS AI platform can easily be tailored to specific needs and requirements.

Figure 2 illustrates the FabOS AI platform service layers.

Figure 2: FabOS AI Platform Service Layers

(No Description)

Open source draws its strength from being easy to integrate and reuse. Consequently, FabOS is not starting from scratch, but integrates and adapts existing solutions.

For example, Eclipse BaSyx will be an integral part of the FabOS ecosystem. Eclipse BaSyx is developed as part of BaSys 4.2, the continuation of the BaSys 4.0 research project. It implements key Industry 4.0 concepts, such as the AAS, and provides off-the-shelf components that enable quick setup of an Industry 4.0 infrastructure.

The FabOS project team also plans to integrate open source solutions from other domains, such as edge computing. That means FabOS will provide an open source infrastructure that is built on other open source components.

Open Communities Drive Innovation

In addition to easy integration and reuse of software, open source also enables an open and transparent process that allows stakeholders, such as users and developers, to be involved. The process is based on global collaboration, distributed change management, and iterative development, along with seminars, workshops, hackathons, and other means of involving FabOS community members.

A process that is this open ensures quality and security because defect-correction cycles and high security assurance can be quickly and effectively implemented.

FabOS uses this transparent approach to implement appropriate security mechanisms and precautions at all necessary levels and to crowd source security requirements from developers and users. Because FabOS is modular and adheres to open source principles, these mechanisms and precautions should always correspond to the state of the art. This keeps the statistical safety and operational reliability (safety) of the applications at the highest possible technical level.

The open architecture also allows for evolutionary designs that fulfill new or changed requirements and are supported by robust and modular components that are interchangeable because they adhere to open standards. Modern and future IT architectures, systems, and infrastructure often consist of modular components that are distributed logically and locally. FabOS is primarily targeting this type of open and flexible architecture, but also offers ways to integrate existing systems and resources.

The open community approach also allows stakeholders to collaborate and to identify new business models. Developers and solution providers can create collaborative distribution models and new services, such as liability support, or solutions for service level agreements. Transparency and participation can also be used as marketing tools, leading to fresh competition, innovation, and customer participation in previously closed oligopolies.

Get More Information

To learn more about FabOS or to get in touch, visit our website.

About the Author

Daniel Stock

Daniel Stock

Daniel Stock is a group leader at the Fraunhofer Institute for Manufacturing Engineering and Automation IPA. His group focusses on research on IT Architectures for Digital Production, which includes building integration solutions for machines, IT infrastructure for production environments and service deployment strategies for edge-cloud platforms for I4.0-compliant applications.

Frank Schnicke

Frank Schnicke

Frank Schnicke is project manager at the Fraunhofer Institute for Experimental Software Engineering IESE. His research focuses on Industry 4.0 software architectures that enable changeability in industrial plants. Additionally, he is coordinating the implementation of the Eclipse BaSyx reference implementation.

This project has received funding from the German Federal Ministry for Economic Affairs and Energy (BMWi)'s AI Innovation Competition under grant agreement No 01MK20010A.