An Algorithm for Variability Identification by Selective Targeting

Large companies have large embedded software systems, where common and reusable software parts are distributed in various interrelated subsystems that also have lots of uncommon and non-reusable parts. The approach finds software parts that may or may not be reusable in a particular application engineering project. It is the task of application engineering to figure out whether the identified components and variants are directly reusable and reuse them in application engineering. In Software Product Lines, the identified reusable common and variable components should be generalized and stored into asset bases. In real life, it may be too much effort and costs to generalize application level assets into domain assets and it is just more feasible to try to find reusable common and variable components directly from existing applications. The proposed approach is selectively targeting the component-feature model instead of an inclusive search to improve the identification. We explore the components and their features from a predefined component node list and the features node vector respectively. Keywords—Design Tools; Embedded Systems; Feature Extraction; Software Reusability; Variability Management.