Variability management support for large-scale software product lines

Due to today's software (dependent) society the amount of software systems as well as their size and complexity increases. Furthermore, quality requirements are raising because software is often deployed in securityrelevant areas as for instance in automotive control systems. To remain competitive, software companies have to make their software production more cost-effective by shortening development cycles. Furthermore, it is a necessity to react to market changes quickly and reduce the time-to-market and thereby improving the products quality. To achieve those goals, practicing software reuse approaches, as for instance product line development, proved to be suitable. However, product line development still exhibits shortcomings in its methodical proceeding as well as the lack of tool support. This thesis attempts to address the issue of rare tool support for software product line development. In particular it focuses on the activity of variability management using decision models. Variability management is an activity of software product line development, which is crucial for the understandability and maintainability of a software product line. To provide a better support for this activity, this thesis identified several approaches to reduce the complexity of decision models in large-scale product lines. The developed concepts were then implemented in a contribution made to the Decision Modeller, an existing variability management tool. Thereby, the emphasis lied on implementation of a graphical modelling support. To evaluate the developed and implemented concepts an experiment was conducted. For the experiment a setup with two groups was selected. Both groups had to accomplish an identical set of tasks. The first group was working with the Decision Modeller without the additional contribution of this thesis, whereas the second group was working with the Decision Modeller and the additional contribution. The experiments result showed that the contribution made to the Decision Modeller substantially helps to reduce the complexity of decision models and thus improves the application of the Decision Modeller in large-scale product lines