Assessing System Architectures: The Canonical Decomposition Fuzzy Comparative Methodology.

摘要

The impacts of decisions made during the selection of the system architecture propagate throughout the entire system lifecycle. The challenge for system architects is to perform a realistic assessment of an inherently ambiguous system concept. Subject matter expert interpretations, intuition, and heuristics are performed quickly and guide system development in the right overall direction, but these methods are subjective and unrepeatable. Traditional analytical assessments dismiss complexity in a system by assuming severability between system components and are intolerant of ambiguity. To be defensible, a suitable methodology must be repeatable, analytically rigorous, and yet tolerant of ambiguity. The hypothesis for this research is that an architecture assessment methodology capable of achieving these objectives is possible by drawing on the strengths of existing approaches while addressing their collective weaknesses. The proposed methodology is the Canonical Decomposition Fuzzy Comparative approach. The theoretical foundations of this methodology are developed and tested through the assessment of three physical architectures for a peer-to-peer wireless network. An extensible modeling framework is established to decompose high-level system attributes into technical performance measures suitable for analysis via computational modeling. Canonical design primitives are used to assess antenna performance in the form of a comparative analysis between the baseline free space gain patterns and the installed gain patterns. Finally, a fuzzy inference system is used to interpret the comparative feature set and offer a numerical assessment. The results of this experiment support the hypothesis that the proposed methodology is well suited for exposing integration sensitivity and assessing coupled performance in physical architecture concepts.