Can tools reduce the costs of developing control engineering-based self-adaptive systems?

Many self-adaptive techniques primarily based on control engineering methods have been proposed for different software system in the literature. These works only evaluate the self-adaptive capabilities of the proposed control solution, but no evaluation is performed to quantify the costs of implementing such a control solution. This paper provides details of an empirical study and its results, conducted to quantify the implementation, testing and knowledge requirement costs when building self-adaptive software systems using control engineering methods. Our objective is to investigate, whether these costs can be significantly reduced if the supporting tools are available. In order to achieve this objective, an empirical study is conducted with a group of software engineers. The findings indicate that the aforementioned costs can be significantly reduced when supporting tools are available. We also list the lessons learned from this study and recommendations, which may be useful in designing similar experiment and improve the validation of selfadaptive techniques in the future.