A tool for modeling and analysis of relationships among feature model views

Abstract

High-tech systems development involves several models and artifacts, each focusing on one or more aspects or parts of the system. In product lines, managing the common and variable characteristics of these development artifacts typically involves feature models. Yet for high-tech systems, management is complicated by the large number of features and constraints. Using multiple feature models, each with a reduced number of features relevant only for specific artifacts, has been identified as a possible solution to deal with this complexity. However, to use multiple feature models effectively, the relationships among feature models need to be known. While several approaches that support multiple feature models exist, they do not explicitly provide concepts for representing structural relationships between features in different feature models. Explicitly documenting these relationships improves understanding of the complete variability model and supports maintainability of the variability model under evolution. In this study, we present two domain specific languages---one for relationship modeling, one for change analysis which enables users to express change impact analysis queries using the relationship modeling language terminology---, which we implemented in TReMAFeM, a tool for relationship modeling among feature models. Moreover, we demonstrate the use of the tool on the industry-related case of the Extended Twilight System product line and explain how it can be used to support typical change impact analysis scenarios.