A framework for agile system modeling

Agile system modeling is a topic that I will also be discussing at the ReConf 2018 will present. However, an important question is how agile MBSE can be practiced. I base this on primarily on the work of Schindel and Dovewho presented their research findings at INCOSE.

I think agile MBSE is a promising approach that is already being practiced to some extent and will gain momentum in the coming years. But there are also skeptics. For example Ronald Carson fundamentally doubts agile systems engineering - whether with or without modeling.

Dealing with risks and changes

Why agility in SE is a good idea, I have already dealt with. In short, it is about reducing risks in the context of change. This primarily refers to economic risks (faster to market with better products). However, this also implicitly includes functional safety and security, because without these, the product will hardly sell.

A central aspect of agile development is continuous learning and dealing with changes that inevitably result from new insights. Models help to deal effectively with changes. Hence agile system modeling, which I have already dealt with in the past.

Information instead of process

Many readers have probably already been in the following situation: the prescribed processes were followed, and yet the result was flawed. This situation illustrates why it is so important to rethink processes in terms of information.

Systems engineering has a strong focus on processes. This can be seen, for example, in ISO/IEC 15288, on which the INCOSE Manual based. Schindel has investigated this phenomenon and found that this is generally not the case in other disciplines.

Processes ensure consistent results, but the results must also be accessible and, above all, useful. This is why Schindel and Dove focus on information. To do this, they use the Agile Architecture Pattern (AAP), which Dove developed with LaBarge.

Learning - and retaining what you have learned!

Information is therefore essential, and new information is discovered during development. This is also known as "learning", and what has been learned must be retained. This is where it gets interesting, because this is where models help. Models make it possible not only to retain what has been learned, but also to retrieve it when it is needed.

Information appears on several levels, and therefore one model is not enough: Here, Schindel and Dove distinguish between the system to be developed (target system), and the development system (system of innovation). These must be able to react to changes, which is shown conceptually in the following image.

Image source: Shingle and Dove

Develop agile systems in an agile way

We can now see that the system to be developed is closely linked to the development model. In this approach, the following aspects are particularly interesting:

• The framework is so generic that it can map both agile and non-agile development systems. It is therefore first and foremost a framework that enables the juxtaposition of different approaches.
• As a framework, an important aspect is the clearly defined demarcation between the systems. Neither is evident from the informal picture. However, Schindler and Dove have also created a formal domain model, which is illustrated in the paper cited above.

Conclusion

This article only describes the most important concepts of agile system modeling. The reader should take away that (1) MBSE supports agile working, that (2) two agile systems must be considered (the system to be developed and the development system) and (3) that there is a framework with which this can be mapped.

Finally, it should be mentioned that the paper by Schindler and Dove contains an appendix that applies the framework to existing patterns, such as the ISO 15288 V-modelfor example. All in all, an exciting approach that has the potential to accelerate the transformation of system development - both in terms of agility and modeling.

Cover picture: Photo by Tony Webster on Unsplash