Supporting architectural decision making for systems-of-systems design under uncertainty

Abstract

For the design and integration of complex systems-of-systems, various architectural decisions for recurring design problems need to be made. This requires that the software architects consider various design issues and alternatives, make trade-offis for competing requirements, and adapt the decisions to spe- cific technologies and systems. Documentations of reusable architectural design decisions (ADDs), e.g., pattern-based de- cisions, provide rather informal guidelines for making recur- ring ADDs. These and other factors introduce many sources of uncertainty in the architectural decision making process. Existing approaches do not consider this inherent uncertainty of architectural decision making, which has been until now largely ad hoc and informal, without explicit, automated sup- port. Apart from that, the design rationale for repeated ADDs often remains undocumented, leading to loss of architectural knowledge. To address these problems we propose to provide semi-Automated support for decision making and documenta- Tion of reusable ADDs under uncertainty using a fuzzy logic expert system. We motivate our approach using a systems-of- systems example from the industry automation area in which our approach has been applied.