5 things we need for an MBSE ecosystem: Answers from the MESConf Open Space

In the 1990s, people complained that software development was inefficient and expensive. The wheel was reinvented every time. This has now changed drastically, thanks to repositories, package managers, semantic versioning, open source and much more.

We had hoped for a similar development at MBSE, but so far it has failed to materialize. We don't even have a functioning model exchange across tool boundaries, let alone across team or company boundaries.

At MESConf, I co-hosted an OpenSpace with Jan Richter, where practitioners and experts explored why we don't yet have an MBSE ecosystem and what we need to create one.

Ecosystem for software

When I attended programming lectures at university in the 1990s, it was normal to develop basic data structures for each project from scratch: Hashmaps, linked list, etc. Today, such data structures are available in the form of standard libraries in every programming language.

And we can also use more complex functions via external libraries. While I still had to enter these manually in the 2000s, today we get them from repositories, which can be public or private. Thanks to semantic versioning and package management, there are hardly any problems with updates.

Open Source and free tools were also important drivers for a vibrant ecosystem for software development. Both enable interested parties to start working immediately and minimize friction losses.

But the whole thing also works because there are functioning business models. This is not a matter of course. When IBM released the Eclipse source code as open source in 2001, experts considered this decision to be very risky. I myself was using a IDEwhich cost about $1,000 per license. In 2002 we replaced it with Eclipse and found the performance better than that of the commercial IDE.

MESConf

I already moderated a session at MESConf last year, on the subject of AI and MBSE. MESConf has been taking place on the infineon campus in Munich for ten years now. As the number of participants is limited, an intensive exchange with top-class participants from the MBSE scene takes place here. MESConf is no longer a secret insider tip.

MBSE ecosystem: Key questions for MESConf Open Space

As related topics are typically grouped together in Open Space, I moderated the session together with Jan Richterwho proposed the topic of multidisciplinary component description. Robust components are an integral part of a functioning ecosystem.

In order to give the discussion the right direction, we asked 3+1 key questions:

1. Why do software ecosystems work?
2. Why do MBSE ecosystems not (yet) work?
3. What needs to happen for MBSE ecosystems to work?
4. What role do multidisciplinary component descriptions play in this?

During the 45-minute session, we discussed these questions and collected answers.

5 things we need for an MBSE ecosystem

At the end of the session, we evaluated the points found by simple voting. The exact results can be seen in the flipchart photos below. Five points were rated as relevant by ten or more participants:

Packaging is one of the reasons why we have such a well-functioning ecosystem in software development. It took us two decades to understand how to realize robust reuse of software components with semantic versioning and package managers.

Incidentally, we have long had a functioning reuse of electrical components (power connections, integrated circuits (ICs)), mechanical components (screws, gears, etc.) and now also complex CAD models that act as a template for 3D printing. However, integration usually has to be carried out manually and the mechanical/electrical/software trio is not yet a robust unit.

In analogy to the PDF format, the participants saw a standard viewer for models as essential. The majority of stakeholders are readers. Giving them access to the model easily and without friction losses is laborious today. Significant in this respect was a presentation by Airbus at the ReConf this year. Airbus mentioned there that the high tooling costs were one of the reasons why MBSE models are not automatically accessible to all employees.

However, a viewer is just one of many tools that we need in MBSE. There is still a considerable lack of interoperability here. This also applies to the use of libraries, if they are used systematically at all. Much here is reminiscent of the situation in software development in the 1990s.

We have taken the topic of bidirectionality from the previous session into this one, where the focus was on the Textual notation of SysML v2 went. One of the questions was whether we should see the textual notation (or also the graphical one) as a read-only view of the model. The group's clear answer was: No! We must also be able to edit in each view, changes must be synchronized and must not become local, possibly outdated copies.

Finally, we discussed the topic of methodology, which is closely linked to tools and language in a triangular relationship. is linked. Although we had not found any clear answers here (also due to a lack of time), we agreed that a proliferation of methodologies also stands in the way of a homogeneous ecosystem.

Flipchart of the session

Conclusion

This MESConf OpenSpace on the MBSE ecosystem was just one of many exciting discussions in contributions to the conference. I'm already looking forward to next year.