A methodological framework for interface requirements identification and specification

摘要

This dissertation develops a methodological framework for the identification and specification of internal, static system interfaces. This interface specification is defined earlier in the development life cycle than is currently proposed by systems engineering standards. The method also incorporates an approach to analyzing component and interface design autonomy by tracking system element definition or design maturity. This maturity heuristic aids in the identification of system element design constraints created by interface dependencies. By implementing the method as an iterative activity between requirements engineering and initial design, an interface specification can be developed that is more complete and accurate than current interface identification and specification methods. The technical approach discussed relies on an enhanced N-Square Diagramming technique and newly developed interface analysis algorithms. Property matching at both ends of system interfaces, using a consistency model, identifies interface specification validity or discrepancy. Interfaces that lend themselves to flow properties are validated by converting the N-Square Diagram into a directed graph and properties are validated by converting the N-Square Diagram into a directed graph and tracing the flow interfaces, through the system to identify missing or discrepant interfaces. Component and interface maturities are analyzed to determine a design autonomy heuristic, providing a relative measure of component and interface design autonomy. The methodology was successfully implemented in a multiple case study which developed interface specifications and design autonomy reports for two systems: the DISN Communication Server and an object-oriented Vector Graphics Engine, demonstrating that the methodology can be used in various domains. The evaluation of the method's effectiveness by five Subject Matter Experts showed that the method achieved anticipated results. This research represents significant contributions to systems engineering knowledge in the area of interface identification and specification. It demonstrated that a methodology can be incorporated such that interface requirements and specifications can be derived earlier in the development life cycle and that the specifications can be more complete and accurate than current approaches. It has also developed a framework for interface analysis using an enhanced N2 diagramming technique with an underlying database and model base. Finally, the research has demonstrated that definition and design maturity can be analyzed to determine component and interface design autonomies. This research represents a first step in developing a rigorous interface identification and specification methodology. Building on these promising results, future research directions are suggested.