Structuring automotive product lines and feature models: an exploratory study at Opel. Issue 1 (March 2017)

• Record Type: Journal Article
• Authors: Oliinyk, Olesia; Petersen, Kai; Schoelzke, Manfred; Becker, Martin; Schneickert, Soeren
• Is Part Of: Requirements engineering. Volume 22:Issue 1(2017)
• Page Start: 105
• Page End: 135
• Publication Date: 2017-03
• Subjects: Feature modeling -- Variability modeling -- Product line engineering -- Case study -- Empirical -- Automotive; Requirements engineering -- Periodicals; Software engineering -- Periodicals; Computer software -- Development -- Periodicals; Logiciels -- Développement -- Périodiques; Génie logiciel -- Périodiques; 005.12
• DOI: 10.1007/s00766-015-0237-z
• Languages: English
• ISSNs: 0947-3602

Abstract

Abstract Automotive systems are highly complex and customized systems containing a vast amount of variability. Feature modeling plays a key role in customization. Empirical evidence through industry application, and in particular methodological guidance of how to structure automotive product lines and their feature models is needed. The overall aim of this work is to provide guidance to practitioners how to structure automotive product lines and their feature models, understanding strengths and weaknesses of alternative structures. The research was conducted in three phases. In the first phase, the context situation was understood using interviews and workshops. In the second phase, possible structures of product lines and feature models were evaluated based on industry feedback collected in workshops. In the third phase, the structures were implemented in the tool GEARS and practitioner feedback was collected. One key challenge was the unavailability of structuring guidelines, which was the focus of this research. The structures considered most suitable for the automotive product line were multiple product lines with modular decomposition. The structures most suitable for the feature model were functional decomposition, using context variability, models corresponding to assets, and feature categories. Other structures have been discarded, and the rationales have been presented. It was possible to support the most suitable structures with the commercial tool GEARS. The implementation in GEARS and the feedback from the practitioners provide early indications for the potential usefulness of the structures and the tool implementation.