Software Product Line Evolution with Cardinality-Based Feature Models

Abstract

Feature models are widely used for modelling variability present in a Software Product Line family. We propose using cardinality-based feature models and clonable features to model and manage the evolution of the structural variability present in pervasive systems, composed by a large variety of heterogeneous devices. The use of clonable features increases the expressiveness of feature models, but also greatly increases the complexity of the resulting configurations. So, supporting the evolution of product configurations becomes an intractable task to do it manually. In this paper, we propose a model driven development process to propagate changes made in an evolved feature model, into existing configurations. Furthermore, our process allows us to calculate the effort needed to perform the evolution changes in the customized products. To do this, we have defined two operators, one to calculate the differences between two configurations and another to create a new configuration from a previous one. Finally, we validate our approach, showing that by using our tool support we can generate new configurations for a family of products with thousands of cloned features.