By Dean ~ December 27th, 2008. Filed under: Modeling, Systems Engr..
It presents an excellent case study of applying architecture centered Model Based System Engineering (MBSE) to the design of an extremely large, complex and distributed project (The E-10A Multi-Sensor Command and Control aircraft which will provide the U.S. Air Force’s next generation airborne Command and Control Intelligence Surveillance and Reconnaissance capability).
The authors do a wonderful job of describing how MBSE was central to addressing the “social” challenges in coordinating design and development across a broad range of constituencies and a long-term projected lifecycle. Their discussion of project management, communication and cross-functional collaboration is extremely compelling.
“… there was a significant need for common mechanisms to allow a variety of people to reason about the system. Associated with this is the need to achieve common understanding of key technical decisions by the project team.”
“To [sic] often, requirements and design artifacts are loosely structured textual statements or graphics (e.g., briefing charts) where there is minimal computer-based enforcement of consistency available. The cost of manual enforcement (not to mention that this sort of work is not something people tend to like to do – and therefore don’t do particularly well) grows as the amount of requirements and design data increase.”
“Because MBSE tends to make information more consistent and readily available (e.g., the model is the design, not hundreds of seperate [sic] textual documents or briefing charts), we believe it was a major factor in an overall high and productive level of collaboration across the team.
We were pleased with the early insight we achieved in understanding system dynamics and also the ability to assess the design’s ability to meet performance and timing requirements.”
While the foundation for their MBSE effort (Executable UML augmented by the UML Profile for Schedulability, Performance, and Time) seems to have met some of their needs, …
“…experience has been that it is often not until the integration and testing phases that issues with things like system dynamics, performance, and subsystem/component interfaces are discovered. The cost of rework at that point can be prohibitive”
We certainly believe that they would have been very well served by employing RAMS methodology to make optimal decisions regarding platform architecture and hardware / software partitioning as early as possible in the design. We feel especially strong on this point, since they chose to use products from some of our competitors for parts of this effort. Using Foresight tools for both modeling (and intent capture), as well as analysis, could have eliminated the potentially risky model transformation described in the paper.
Other than that unfortunate oversight, this is an excellent presentation of MBSE’s ability to foster high bandwidth communication across the full spectrum of project stakeholders as well as to accelerate project cycles by doing better design. If you’re at all interested in applying Model Based / Driven System Engineering to your complex design tasks, this document is worth a look.