Modeling Relationships Between Feature Model Views

The development of high-tech systems involves several models and artifacts, each focusing on one or more aspects or parts of the system. In product lines of such systems, managing the common and variable characteristics of these development artifacts by means of feature models gets complex due to the large number of features and constraints. Using multiple feature models, each with reduced number of features relevant only for specific artifacts, has been identified as a possible solution to deal with this complexity. However, the advantages of explicit representation of relationships between multiple feature models remain unexploited. These relationships need to be represented and managed explicitly in order to improve understanding of the complete variability model and support maintainability of the variability model under evolution. In this paper, we propose a modeling language that can be overlaid, non-intrusively, on existing multiple feature models to express the relationships between feature models. The language uses the concept of views to represent different aspects of the product line variability model, each view corresponding to a feature model represents variability information in one or more development artifacts of the product line. We give a metamodel for relationships between feature model views and provide the semantics of the relationships. We present how proposed relationship types can be used to analyze the impact of a change on the complete variability model. Furthermore, we provide formal definitions of relationship types, making them suitable for existing automated analysis operations in the literature.