An Approach to Detect the Origin and Distribution of Software Defects in an Evolving Cyber-Physical System

Cyber-Physical Systems (CPS) are usually developed by an incremental approach. A changing environment like demanding user requirements or legislation amendments lead often to multiple development paths in an evolving CPS. Hence, software variability plays an increasingly important role adapting the characteristics of such CPS to different contexts. This paper focuses on software variability realized through a Software Product Line (SPL) more specifically. Thereby, variability and evolution are usually managed in different tools. However with respect to software defects, a holistic handling of variability and evolution is necessary to ensure a reliable software defect removal. Particularly, detecting software defects in different evolution stages and derived variants is ordinary, but complex and error-prone. To close the gap between variability and evolution, this paper presents a systematic approach to combine both disciplines. In particular, we apply existing variant management techniques in combination with software configuration management methods to determine a software defect's origin and distribution in an evolving SPL. We apply our approach to a CPS from the automotive domain to show its industrial relevance and usefulness.