5 articles on SysML in the INCOSE Special Issue
INCOSE has published a special on SysML, which can be read free of charge at Wiley until April 15, 2022. After that, it is subject to a charge or requires a membership.
First things first: Here is the Link to the Special Issuewhich consists of five papers. The following is a summary of the publications:
Employing SysML to model and explore levels-of-service: The case of passenger comfort in railway transportation systems
SysML can do a lot - but not everything. For example, one obvious option is missing, Traffic quality (Level of Service, LoS) for transportation systems. The authors use SysML profiling to solve this problem. To this end, they create an implementation framework that enables the definition of LoS parameters. These are based on system requirements, LoS requirements and estimates of the expected LoS. Compliance with the requirements can then be checked and optimized using simulations. The focus was on passenger comfort as the main performance indicator.
The paper also includes a case study, which the authors apply to the metro in Athens. Using data provided by the metro, passenger comfort was determined (e.g. Line 1 has LoS "C"). Using simulations, the team was then able to run through various scenarios, such as higher frequencies and train cancellations.
Verifying SysML activity diagrams using formal transformation to Petri nets
There are a number of modeling languages. Some are more formal, others less so. Formal languages enable analyses that cannot be realized with less formal methods, for example mathematical reasoning. SysML is sometimes referred to as "semi-formal". The authors of this paper describe a way to translate behavior modeled in SysML into Petri nets in order to be able to use the formalism of Petri nets.
I find this direction of research highly interesting, as it combines the possibilities of formal modeling with the acceptance of SysML. We get the best of both worlds.
As expected, a transformation is only possible with restrictions. The authors have drawn up a catalog of transformation rules. The devil is in the detail, as can be seen in the following example. Furthermore, the authors have also discovered different, equivalent transformations.
Systems modeling language extension to support modeling of human-agent teams
We develop for people. Therefore, the interaction of people with systems is one of the most important aspects of system development. To this end, the authors have proposed an extension of SysML to describe and capture the structure and assumed interaction of human-agent teams. Part of the work is a corresponding methodology. First, the concepts of the system are captured. The methodology then enables the derivation of requirements, structure and high-level activities that can serve as the basis for system-specific system models.
I am a little skeptical about how much added value this approach really brings in practice. The authors also admit that they need to develop the approach further because it is incomplete. Nevertheless, the need for a systematic analysis of human system interaction is beyond question. I therefore welcome research in this direction.
Model-driven architecture based security analysis
Attack security has been an important topic for years. The authors propose a model-based approach to security analysis that consists of three steps. In the first two steps, both the security requirements and the system are modeled conceptually. In the third step, these models are transformed for verification. SysML profiling is used for this.
A new profile enables formal and temporary modeling of attacks Another profile enables the transfer of data streams between different, mostly heterogeneous communication subsystems.
Users can further utilize the models via transformations, for example for checking via Modelchecker (NuSMV).
What I find almost more important is that this approach ensures that attack security is incorporated into the architecture at an early stage. The automation potential of the approach also supports this.
Comparative analysis of a model-based systems engineering approach to a traditional systems engineering approach for architecting a robotic space system through knowledge categorization
I personally did not find this last paper extremely innovative, but important. For readers of this blog, it is obvious that MBSE is a sensible approach above a certain level of system complexity. But in many companies, decision-makers still need to be convinced that MBSE is worthwhile. This paper supports decision-makers in this argumentation.
I would describe the authors' approach as "classic"; they create a framework for the architecture based on MBSAP, MagicGrid and STRATA. This covers the three organizational levels of the system as well as the structural, behavioral, data and requirements perspectives. The authors compared this approach with the document-based approach using a case study.
The authors classified the artifacts of both approaches using the Revised Bloom's Taxonomy, which enabled a quantitative comparison. Not surprisingly, the authors were able to prove that MBSE captures architectural knowledge more completely than the document-based approach, as can be seen from this table:
However, I am not entirely satisfied with this conclusion for two reasons: Firstly, this is a hypothetical case study. It would be much more convincing if the system to be developed were actually built.
Comparing MBSE with Microsoft Office is not fair or realistic.
I find it even more problematic that the authors have compared MBSE with Microsoft Office (Word, Excel, PowerPoint). Because there are many options between these two extremes. In particular, modern requirements management systems enable a Primitive entity-relationship modeling. Although this is a far cry from MBSE, it has extremely high benefits.
Conclusion
This special on SysML from INCOSE. Too bad that the content is only available until April 15.
Photo by Kelly Sikkema on Unsplash
