Organizations acquiring new systems often pay for the construction of the same components (design, documents, and code) over and over again. The historical premise of reusing existing components to save money, reduce risks, and improve schedule performance has so far proven more myth than reality. The technology customarily used to support reusability is a library of reusable components. Results from this kind of library have been disappointing in the real-time simulation realm. This library usually has no mechanism for showing interrelationships or structure of components. And as the complexity of components increases beyond control-law fragments, the components become less and less reusable and the size and contents of the library grow out of control, until a potential reuser finds that using the library is more work than rewriting a component from scratch.
The DARPA Software Technology for Adaptable, Reliable Systems (STARS) program is addressing these and other problems of reuse, and promoting technologies to achieve a vision of process-driven, domain-specific, reuse-based, technology-supported system development.
The Software Productivity Consortium (SPC), of which Boeing is a member, is pursuing a vision of transforming the production of software into an engineering discipline. One result has been the development by the Consortium of the Synthesis methodology, an innovative reuse-oriented approach to software development. Synthesis defines how, guided by the commonalities and variabilities of a family of systems, to create adaptable components and a process for choosing among the variabilities to create systems. Because this methodology exemplifies the major tenets of the STARS reuse vision, Synthesis is under investigation by STARS and Boeing.
STARS has funded a pilot effort by Boeing Simulation and Training Systems, the Center of Excellence for simulation and training for the Boeing Company, to use and evaluate the SPC Synthesis process. This project will determine variability information and elaborate a decision process for one domain of a generic flight simulator and test the decision process on an existing F-16 R&D simulator. This paper reports on the research performed and the preliminary results obtained.