Digitalization has been the central topic in production technology for many years. On the one hand, it is used to pursue traditional goals such as productivity and precision, while on the other hand, digital added value such as predictive maintenance, AI-based approaches and traceability are increasingly being realized. The development processes and the view of the overall system are traditionally associated with a focus on mechanics and electrics; digital services are typically seen as an add-on and developed as concrete solutions for a specific task. While some of these solutions successfully make it into application, many approaches fail on their way into industrial practice. The reasons for this are a lack of scalability, high costs, integration problems and complex engineering. The frequently invoked "value of data" is often negated by the "cost of data". People often look enviously at the innovative dynamism of other sectors. The IT world is seen as a successful alternative with its high innovative strength, which is characterized by adaptability, fast innovation cycles and a flexible response to customer requirements.

Figure 1: Software-defined manufacturing - continuity and consistency based on an ecosystem.

© ISW

Successfully digitized industries, such as mobile communication, show a structural difference: Digitalization is not used to implement individual solutions. Instead, it creates an ecosystem in which solutions are developed on the basis of the available resources. Such an ecosystem goes far beyond a purely technical system: methods, processes, business models and, in particular, the mindset are integral components.

Establishing such an ecosystem is the focus of software-defined manufacturing (SDM). Figure 1 combines the central elements of this approach: production technology applications are implemented on the basis of resources that are abstracted using models and interfaces. Instead of following a linear development method, engineering, operation and optimization represent a continuous process that is implemented on the basis of consistent data models. In order to take into account the more critical requirements of production technology, new solutions are first tested against digital twins and not directly in the field, as is usual in the software environment.

A digital ecosystem

Today's production systems are usually relatively rigid, often optimized specifically for a product to be manufactured, hardware and software are tightly coupled, a large number of proprietary solutions are used and the coupling with IT systems is limited. This contrasts with the ecosystem-based approach. With this approach, various modules are available as resources. They perform specific tasks and interact via standardized interfaces. They can also be re-orchestrated using software engineering methods to fulfill changed production tasks. They are also consistent and directly linked to IT systems.

But how can such an ecosystem be created for production technology? Although the SDM approach differs significantly from the state of the art, "software-defined" itself is widespread: Software-defined networking (SDN), the entire IT world and every smartphone implement this idea. They all combine software, hardware and connectivity.

Figure 2: Simplified SDM architecture: solutions and resources are linked by data and services.

© ISW

However, trying to develop an ecosystem for SDM using the classic approach of requirements, specification and implementation would be doomed to failure. Instead, the development processes of the examples mentioned above must serve as a model. Iterative development, which is driven by both applications and technology, must take place in parallel with the establishment of the ecosystem. From a technological perspective, a large number of suitable solutions are currently being developed, for example in the context of real-time capable communication using standard technologies or deterministic compute solutions in processor architectures. These approaches need to be consolidated and merged and, if necessary, expanded.

The key to establishing the ecosystem is the industry-wide development of a common perspective and understanding of the system. To this end, an architecture is defined within the framework of SDM, which basically defines SDM solutions based on resources. Resources must be mappable using suitable models and software-definable using suitable integrated or external services. Solutions and resources are linked via the SDM core, in which all information and models are combined in a data space and basic services - such as orchestration - are available. Figure 2 shows the architecture in simplified form.

SDM solutions can be classic production tasks, digital added value, but also an optimizing function. On the resource side, in addition to classic OT systems - such as sensors and actuators - there are also purely digital resources - such as virtual controllers (vPLC) - as well as infrastructure, i.e. computing and communication resources. All of these resources need to be virtualized: for example, platforms using virtual machines or container technologies.

The necessary connectivity

Figure 3: The dimensions of convergence.

© ISW

Connectivity is a fundamental prerequisite for a software-defined ecosystem. This means that all resources, from the virtualized platform and physical components to the digital service, must be identifiable and configurable. But they must also be able to communicate with each other, with central services and with internal and external stakeholders. This connectivity must satisfy the various framework conditions of the resources, demonstrate a high degree of flexibility and interoperability and also fulfill deterministic requirements.

It is therefore essential to rely on interoperable IT technologies that can meet the deterministic requirements using Time-Sensitive Networking (TSN): Ethernet, wireless connections with deterministic Wi-Fi and 5G, as well as DetNet for connectivity via routed networks. Various technologies are used at the upper layers of communication. OPC UA is seen as having great potential in this context. The result is a network that is convergent in terms of both data classes and technologies (as shown in Figure 3 ).

Convergent communication as a resource

From the perspective of the SDM ecosystem, communication technology is a resource that can fulfill various tasks: On the one hand, it is responsible for basic connectivity, but on the other hand, it can also provide a limited number of deterministic connections. In order for this resource to be used, it must be digitally mappable in suitable models. On the other hand, it must also be configurable.

A large number of domain-specific models for devices, topologies, configurations and also for describing specific data traffic already exist for mapping communication. Some of these models are specified by the IEEE or the IETF. In order to integrate them into a holistic description of an SDM ecosystem, they can be referenced using administration shells, for example.

As far as configuration is concerned, the individual technologies also offer a large number of solutions. As a rule, these are standard IT technologies that need to be managed. The central approach is therefore software-defined networking, which can already cover many aspects of connectivity. Various technology-specific solution approaches are currently being developed to utilize the deterministic properties, such as central management for TSN.

Ecosystem: the answer to complexity

Configuration is still seen as a key challenge for TSN technology as a whole. While solutions have been found for closed TSN-based systems, questions remain unanswered for complex systems in an open ecosystem. In particular, multiple, independent applications in combination with virtualized platforms still pose challenges. If not only purely Ethernet-based TSN is considered, but also wireless technologies, for example, the complexity increases further. Opinions differ widely on the extent to which these challenges can be solved and by what means. On the one hand, pragmatic solutions are sought, intelligent tools are developed or the use of complex features is dispensed with.

An ecosystem offers the possibility of a different approach here: the acceptance and utilization of full complexity through encapsulation by means of central services and suitable abstraction layers and models. This basic approach is consistent with other software-defined ecosystems, for example modern mobile phone technologies in the mobile communications sector or complex peripherals via USB in the office sector. Not only the configuration of the communication itself can be solved, but also the integration with related topics such as virtualization.