Differentiating System Architectures: Applying Architecture Measures

Abstract

AbstractHaving the right system architecture can facilitate system development and help manage life‐cycle costs. The wrong system architecture can make these more challenging because there may be missing program activities to address the key features of the selected architecture. For example, integrated architectures may require more design integration and analysis compared with federated (modular) architectures because of the tight coupling among subsystems. In the extreme case, the wrong architecture can lead to failure to satisfy all the requirements or obviate the possibility of future enhancements.How do we select the right system architecture? Can we differentiate a “good” from a “bad” system architecture, or a “better” vs. a “worse” one? When or why should we prefer “modular”, “open”, “service‐oriented”, or “integrated” as an architecture approach? The answers to such questions can help us with the system design of aircraft, space systems, networks, and other complex systems.This paper applies recent work of the National Defense Industrial Association (NDIA) and International Council on Systems Engineering (INCOSE) to evaluate examples of system architectures for different life‐cycle criteria. The paper reviews industry standard definitions of “architecture”, identifies the latest industry work on system architecture measurement, and applies the measurements to specific examples of system architectures. From this analysis an analyst can identify “acceptable” and “better” system architectures. It is shown that there is no universal statement of a “better” architecture because most selection criteria are program‐specific.