The primary goal of this research is to explore new methodologies that support building reliability into the design of software product families. This enables a sufficient formal foundation to compare design alternatives early and throughout the software development. There are several notable challenges, among them the unsuitability of existing reliability modeling approaches for product families due to their inability to handle architectural change and leverage reuse effectively. The relationships and dependencies among products within the family further complicate the matter. The project will: 1) Extend existing (primarily) structural product line architecture modeling and analysis approaches with rich behavioral constructs, and corresponding analyses; 2) develop architecture-based approaches to reliability modeling of software product families; 3) research the interactions between different products reliabilities within a family of software products; 4) develop a decision support system that embodies these principles to compare alternative design choices based on their impact on the reliability of specific products, as well as the reliability of other products within the family. Evaluation of the methods will be performed in collaboration with NASA/JPL to demonstrate broader impacts.
