SysML Lessons Learned

1
SysML Vendor Perspective
Jim Rice jim.rice_at_nomagic.com Technical Sales
Solution Engineer No Magic Inc. www.nomagic.com
2
No Magic Inc.

• Know your panelist
• Founded in 1995 with the vision that there is no
  magic to develop better software.
• Offices
• USA
• Lithuania
• Thailand
• Flagship Product MagicDraw UML
• Industry standards-compliance and support
• Independence from specific development
  methodology
• No environment lock-in (Windows, Linux, Mac)
• No vendor lock-in
• Collaboration and team development
• Core to MDA solutions
• Fast and free customer support

3
No Magic Perspectives
SysML Today and Beyond
4
Integration starts at the SysML Spec and
implementation

• One of the better specs out of the OMG (System
  Engineers were involved)
• No such thing as perfect
• Still needs a little work to reduce confusion
• Needs more and deeper examples - though better
  than most OMG efforts
• Could use and expand on the INCOSE SE Handbook
  examples
• INCOSE/OMG tutorial greatly increases the
  understanding of how and why SysML is to be used

OMG Systems Modeling Language (OMG SysML), V1.0
OMG Available Specification
5
What parts of SysML were hard to implement?

• Is there a System Engineer in the house?
• UML Expert ! SysML Expert
• UML User ! SysML User
• Tool Developer not necessarily a System Engineer
  (or user).

6
Model-Based Engineering Market SysML Customers

• Categories
• Astronomy
• Nuclear Engineering
• Aerospace
• Medical
• Defense
• Software Engineering
• Trends
• Transitioning to SysML with increased
  understanding and support
• Interest in calculations
• Increasing parametrics

7
Interesting Customer SysML Applications

• DODAF and SysML are being mixed to create
  architectural views of System of Systems
• European Extremely Large Telescope
• Hearing instruments for the hearing impaired
• Internally, we are using SysML aspects to define
  product requirements

8
SysML should help reduce failure through valid
and complete models
http//shippai.jst.go.jp/en/Search
9
Why we love SysML

• SysML is an important innovation!
• Brings multidimensional modeling to the SE
• Has the promise to bring SE rigor to other
  disciplines
• Basis for further improvement
• Increases usability of SE designs

10
Positive Feedback

• Improved interdepartmental communications
• E.g. Between business, hardware, and software
  groups
• Requirements validations
• Not fully addressed with use cases
• Tracing requirements to the model and
  satisfaction validation
• Parametrics
• This is also somewhat of a problem as UML models
  are not calculation machines.
• Interface to 3rd party tools better suited to
  this function (E.g. InterCAX)
• Most popular constructs
• Internal Block Diagrams (IBD)
• Activity Diagrams

11
Issues and Concerns

• Validation
• Currently we are conducting bulk validation. Our
  next release of our SysML plugin will support
  real-time validation on the diagram.
• Domain-specific components / types
• Model Libraries
• SysML is a graphical notation and does not
  provide the calculation task and simulation
  required by engineers
• Addressed by vendors by ability to export models
  in standards-based formats for 3rd-party tools.
• Problems with requirements decomposition
• Confusing valueTypes, units, and dimension usage
• Inability to represent configuration or time
  slice values on (IBD)
• Lack of instance models
• Confusing the usage of ports with
  provided/required interfaces

12
Most asked for features

• What are customers looking for?
• Simpler implementation of deep nesting and
  ability to model many configurations
• Integration with simulation tools
• Dont allow users to shoot themselves in the foot
• Model traceability automatically
• Change management

13
Academia Opportunities

• Build and Vet
• Domain-specific components / tests / models
• Industry-specific requirements model
• Satellites (E.g. insulation on a wire)
• Automotive
• MARTE (http//www.omgmarte.org/)
• Critical Review
• Validation of Domain Libraries
• Ensure that any one piece of a model does not
  introduce inadvertent corruption in the rest of
  the model
• E.g. Fuel Injectors, O-Rings, Screws
• Innovation, Evangelism, Education, Partnerships

14
Keys to Accelerated Deployment

• Education
• Seminars / Webinars
• Best Practices in dealing with complexity
• Evangelism
• Spread the word of successful implementations
• Share understanding / change thinking
• Tools Continued Improvement
• If the process for using SysML is too
  complicated, people will not use it (or wont use
  it for long).
• Keep the Real World in mind
• Safety and preservation of life
• Cost / Benefits (ROI)
• Deadlines

15
SysML is just a modeling language, not a solution

• SysML on its own is a start
• Need to mix with other technologies
• Extend via other profiles
• Integrate with other tools math/physics/CAD
• Use MDA to generate code
• Transform to other format
• Common Model Libraries

16
Suggested areas of improvement

• Ensuring SysML Success
• Continued refinement
• Drive UML changes to reduce extensions for SysML
• Certification of model and data exchange
• XMI
• EMF
• Mapping of ISO EXPRESS to UML/XMI
• Semantic model exchange
• Validation of semantics

17
OMG SysML 1.0 and MBSE Coverage

• It is a fairly newly approved specification and
  will need ironing out.
• Real World deadlines require prioritization
  therefore everything can not be included in the
  first release
• Extensibility
• Gaps in specific standards in modeling for
  specific domains
• Lack of standard component pieces (E.g. the
  tracks dont meet)
• Need for continued and increased industry
  participation to ensure acceptance and validity
• Facilitated feedback

18
Where Else?

• Software Engineering
• Emergency Response / Disaster Preparedness
• Logistics
• Civil Engineering
• Financial Systems
• Construction

19
References

• No Magic, Inc.
• http//www.nomagic.com
• http//www.magicdraw.com
• Japan Science and Technology Agency (JST)
• http//shippai.jst.go.jp/en/Search
• OMG SysML website
• http//www.omgsysml.org
• http//www.omgsysml.org/INCOSE-2007-OMG-SysML-Tuto
  rial.pdf
• Refer to current version of SysML specification,
  vendor links, tutorial, and papers

20
References

• PAPERS
• Simulation-Based Design Using SysML Part1 A
  Parametrics Primer
• RS Peak, RM Burkhart, SA Friedenthal, MW Wilson,
  M Bajaj, I Kim
• Simulation-Based Design Using SysML Part 2
  Celebrating Diversity by Example
• RS Peak, RM Burkhart, SA Friedenthal, MW Wilson,
  M Bajaj, I Kim
• The SysML Modelling Language
• M Hause, 2006.
• BOOKS
• Tim Weilkiens. Systems Engineering with
  SysML/UML, Morgan Kaufman Publishers, 2007.
• SUBJECT MATTER EXPERTS
• Daniel Brookshier (danielb_at_nomagic.com)
• Nerijus Jankevicius (nerijus_at_nomagic.com)
• Kritsana Uttamang, Ph.D. (kritsana_u_at_nomagicasia.c
  om)

21
I ? SysML
22
UML Challenges

• UML is not SysML
• As a profile, we needed to add new features to
  support new semantics
• Needed to hide UML
• Needed new capabilities
• Matrix
• Reports
• Notation
• Ease of use for new tasks
• UML is poor at conveying deeply nested assembly
  and configurations so more changes were required

23
The issues with SysML as a profile of UML

• Prior Profile Experience
• Profile is generally an extension of UML that
  uses the same base semantics
• SysML Profile Experience
• SysML is not exactly an extension and remaps
  concepts
• Notation is significantly different in some areas
• Intended usage and process is different
• Level of correctness is greater (and possible)

24
SysML Goodness

• SysML is changing UML modeling
• SysML better than UML for system engineering
• Better ability to show configurations,
  requirements, and systems
• SysML makes UML better

25
How Well Does SysML Integrate With UML?
UML
SysML
26
(No Transcript)
27
SysML in MagicDraw
28
SysML in MagicDraw
29
MagicDraw, Inc.
30
MagicDraw, Inc.
31
MagicDraw, Inc.
32
MagicDraw, Inc.