DOD Considers Virtual Requirements Integration and Agile Methods

DOD reports support the move toward model-based systems engineering. Requirements and specification generation may become both more virtual and agile.
By John Blyler, Professor, PSU – Systems Engineering Program
Below is an useful chart from last year’s (Apr. 2014) review of Model-based Systems Engineering (MBSE) by the DOD. The report reveals the practices and recommendations for the future directions of the DOD in the area of systems engineering.
The current challenges in systems acquisition were summarized as follows:

• Linear acquisition process
• Lack of adaptability to change
• Stove-piped workforce and data sources
• Information shared through static documents
• Limited reuse

How are these challenges to be met? The potential solution was listed as Model-based Systems Engineering (MBSE). “MBSE is part of a long-term trend toward model-centric approaches (in Systems Engineering) adopted by other engineering disciplines, including mechanical, electrical and software (INCOSE SE Vision 2020).” The MBSE approach has the potential to reduce time, cost and risk to develop, deliver, and sustain systems, through the use of models as a fundamental element of program information (see Figure 1). The report acknowledge the multiple efforts to operationalize MBSE through: a)methodologies, tools, languages, standards; b) organizing engineering data into a model for use across a program, and c) implementation within research.

 
 
 
 
 
 
But how exactly will MBSE be used in systems engineering. For example, will requirements be derived from the models? Will there be a merging or sandwiching of traditional textual documents with MBSE models?
This question was further explored in the final draft report to the NDIA Model-based Engineering Subcommittee (2011). In the section on “Virtual integration to Manage Risk throughout the Life Cycle,” the groundwork was developed for the migration from a traditional Waterfall development process to the classic Vee-Diagram model. This approach is also in-line with model-based engineering tools and practices. As noted in the report: “Model-based engineering tools and practices enable an adaption of the classic development “V.” (see Figure 2) Rather than simply decompose the system into its component down the left side of the V and then integrating and testing these components up the right, MBE enables incremental testing of the components during development, well before designs have been committed to hardware. This introduces the ability to manage risk incrementally. At each step in refinement of the detailed design, systems, subsystem, and component requirements can be virtually tested against the current understanding of the design requirements and assumptions. Tested and validated subsystem elements increase the confidence that the system will operate as intended, while tested and validated components elements increase the confidence that the subsystem will operate as intended. Thus testing and validation can be more tightly coupled to the design activities allowing more rapid design convergence and validation of derived requirements in the same phase of the development lifecycle.”

Compare this approach to recent efforts to tailor hardware design to the software development Agile methodology.
The question remains as to the best way to combine past best practices for requirements generation with newer model-based methods. What do you think?