Semantics-Preserving Merging of Feature Models

Large and globally operating enterprises can be confronted with situations where local variability models representing the constraints of individual countries and markets have to be integrated to support a centralized variability management. For example, a car producer operating in the US as well as the European market, could be interested in having a centralized variability (feature) model representing the variability spaces of all supported markets. To achieve this goal, existing local feature models and the corresponding knowledge bases have to be integrated in such a way that the configuration spaces remain the same, for example, for the European market, we would request to support exactly the same set of car configurations that are supported by the corresponding local feature model. In this paper, we introduce an algorithmic approach that supports the merging of feature models in such a way that the semantics of the original feature models is preserved. We present our algorithm and the results of a solver performance analysis which has been conducted on the basis of real-world feature models.