Bridging Model-Based Systems Engineering and Model-Based Design

--sponsoring. Presentation round.

All right, great. Can you see my screen all right?

Yes.

Yes.

OK, fantastic. All right, excellent. Well thank you so much for having us today. My name is Becky Petteys, I am Assistant Engineering Segment Manager here at MathWorks. And it's been great to hear some of these other presentations from some of these other companies, because we are a relative newcomer to the world of MBSE, and so I'm really excited today to be able to present to you some of the stuff that we've been working on in this space.

So just to get started, one of the main challenges that we've heard from engineers like you is that it's very difficult to do some of that conceptual work of systems engineering, and then bring it into the more concrete environment where-- design environment where implementation details need to be specified.

So the idea of System Composer is to help you to bring the ability to do many of these common system engineering tasks into the same environment as you do your-- as you do your design work.

So that's the overall idea. Now why did we start even down this path? Well, so some of the things that we've heard from you, the Simulink and MBD users out there, is that the current use of-- the current tools that they have, they're not especially happy with.

So only 8% of the folks that we've asked said, absolutely I'm happy, I have no problems. The rest had some pains that they were experiencing. So what were those pains?

The number one pain was that a lot of the work that they did in their system's architecture tool, it wasn't executable. Secondly, it wasn't synchronized with their designs, and specifically with Simulink and Matlab in the NBD world. And then too, oftentimes they found it a little bit hard to use, and it wasn't analyzable.

In fact, one customer told us, look, we've tried to build the architectural model into SysML and then connect it to design in Simulink and this just does not work, at least without rework on both the architecture and design worlds whenever a change is needed. This is broken, and we need a more integrated approach.

So how did we try to address these pains? Well, so we really took to heart some of the feedback that we had gotten. And we tried to make a solution that was intuitive, that would help you to tackle the complexity of these real-- these real life systems to allow you to facilitate analysis early and often as you're working through these problems. And ultimately to enable implementation, to seamlessly connect to your design environment.

OK, so let's take a look at how this actually works. So you can quickly and easily you get started Sketch system and interfaces. The nice thing here is that you can start very conceptual and very high level, and then begin to elaborate these things incrementally.

Additionally, you can start to extend some of the basic elements of System Composer, using things that you're probably familiar with, like stereotypes. Where you can find-- and you can find your own custom metadata for these. In addition, this lives right within the same canvas as you can-- as you're working with your architecture in.

Similarly, we have the ability with our tool Simulink Requirements to trace all of these system requirements to these architectural components. Again, this is right in the same canvas as you're working on your architecture models.

So again, once you start to actually get going with this, things start to get really complicated really fast. So this is an actual customer model that we helped them to build. One of the nice capabilities that we have is the ability to create these custom filters that allow you to automatically generate simplified views. Again, only look at the parts of the system that you're really interested in. And this can take a very large and very complex model and make it more easily digestible and communicatable.

Similarly, sometimes you don't need to see all of the connections for a particular model, and instead you can generate something that shows instead the hierarchy of the overall diagram. And no need to do anything extra for this. You can switch up the view, and now you have a hierarchy diagram instead of the flat component diagram.

Now one of the really nice things about sitting on top of a powerful tool like Matlab is that it makes it really nice and easy to be able to analyze your system characteristics. You can quantitatively evaluate your design decisions.

And using not just Matlab, but other toolboxes. For example, Optimization. This allows you to construct some pretty complex and pretty powerful ways to assess. For example, to do your trade studies and assess different design decisions.

Also, as some of the other presenters this evening have talked about, you typically have multiple architectures to describe the same system. You need to be able to show the relationships between these different architectures.

So for example, I might have a functional decomposition for my system, and then also a logical architecture. It's a really important step to be able to allocate functions to these logical components. And you can absolutely also do that with System Composer.

We have an allocation editor that allows you, for example in this case, to put the functional decomposition along the rows, and the logical architecture around the columns, and then quickly add allocations just by double clicking each of those squares.

Also, again, we sit on top of Matlab, which is great, and it allows you to analyze different allocations in order to see quantitatively which allocations actually meet your requirements. And you can do this using the properties that you've defined in your stereotypes.

In our 2021a release, which just came out in March of this year, we've added a couple of new ways for you to be able to define behaviors of architectures and components. So from your architecture, you can, for example, create a sequence diagram. A sequence diagram, as you all probably know, describes the desired sequence of messages that gets passed between components.

But now, I built up the sequence diagram from my architecture, but let's say I want to add a lifeline to my sequence diagram that doesn't have a corresponding component in my architecture. That's not a problem. We will actually automatically add that component to you to ensure that your architecture and your sequence diagrams stay in sync.

Similarly, if I add a new connection between two components in my sequence diagram that doesn't yet exist in my architecture, we'll add that as well.

However, now let's say we delete a component in our architecture. Are we automatically going to delete that from your sequence diagram? No, we're not going to delete things. But we will flag that something has changed for you, that you need to fix up.

So we'll add for you, but we will never take things away. We'll let you instead look at the highlighted elements and then figure out what is the right thing to do there.

OK, the other new way to define behaviors in our latest release is that now you can directly define state charts inside of a component in your architecture. So telling System Composer to create a Stateflow chart behavior will open up an empty state chart with the interfaces already populated from that system component.

So this interface will always be synchronized with that system component. Any changes you make in either direction will propagate. So then you can fill out the behavior of the state chart using our popular product Stateflow.

All right, and you can always shift seamlessly into model based design by linking Simulink models into your system architectures and using the power of simulation to analyze and evaluate that system. So you can absolutely start to simulate your architectures as soon as you have enough behaviors defined, either through Simulink or the state charts I just showed.

Becky, sorry to interrupt. About a minute or so left. Thank you.

Excellent, that's perfect. OK, great. So what can System Composer-- hopefully I've shown you today-- this is an excellent time to wrap up, and I shall do so.

As a whole, I have shown you today some of the things that the System Composer can do. So it makes it very easy to get started. We've really strived to make it easy to get going without the steep learning curve. It can scale up to realistically large systems, and we've demonstrated this with real customer examples.

We can start to incorporate analysis almost directly from the beginning, which is fantastic. Again, we have the benefit of sitting on top of Matlab. And again, it provides an easy on ramp to model based design, directly from the system tool that you're working in.

Just a quick example of how folks have actually already been using this. We've been working with a great group at Gulfstream, and we actually wrote a joint paper for the Digital AVX Systems Conference back in October of last year. And you can get the paper from IEEE.

But really, this was a great quote from one of our champions there, Chris Watkins, that, the goals of System Composer are to make system modeling easy, flexible and scalable, and to ease the transition to the design environment. And the features and constructs of System Composer reflect the prioritization of these goals. That was a great thing to hear from one of our customers.

So hopefully I've shown you today this powerful capability of being able to bridge between model based systems engineering and model based design in the same environment. That's our vision, and our goal for you, our customers. And so we would love to work with you to help you to achieve this vision.

All right, thank you so much for your time.