Applying Hierarchical Feature Modeling in Automotive Industry

Abstract: Context. Variability management (VM) in automotive domain is a promising solution to reduction of complexity. Feature modeling, as starting point of VM, deals with analysis and representation of available features in terms of commonalities and variabilities. The work is done in the context of an automotive industry – Adam Opel AG. Objectives. This work studies the automotive specific problems with respect to feature modeling, investigates what decomposition and structuring approaches exist in literature, and which one of them satisfies the industrial requirements. The approach to feature modeling is synthesized, evaluated and documented. Methods. In this work a case study with survey and literature review is performed. Survey uses semi structured interview and workshops as data collection methods. Systematic review includes articles from Compendex, Inspec, IEEE Xplore, ACM Digital Library, Science Direct and Engineering Village. Approach selection is based on mapping requirements against discovered approaches and discussion with industry practitioner on the regular meetings. Evaluation is proposed according to Goal Question Metric paradigm. Results. The approach that can be followed in the case organization is described and evaluated. The reasoning behind feature modeling approach construction and selection can be generalized for other organizations as well. Conclusions. We conclude that there is no perfect approach that would solve all the problems connected to automotive software. However, structuring approaches can be complementary and while combining give a good results. Tool support that integrates into the whole development cycle is important, as the amount of information cannot be processed using simple feature modeling tools. There is a need for further investigation in both directions – tool support and structuring approaches. The tactics that are proposed here should be introduced in organizations and formally evaluated.