Challenges of testing self-adaptive systems

Abstract

Self-adaptive systems can change their behavior due to internal or external issues detected during operation. Such systems should be able to change their internal structure or functionality to cope with broken motors or changes in the infrastructure. Assuring that the adaptations taken during operation do not impact the desired behavior or functionality of the system is of uttermost interest. In this paper, we contribute to the corresponding quality assurance challenge. In particular, we focus on a specific class of self-adaptive systems utilizing health states for computing repair actions. We discuss the requirements of testing methodologies for such systems and raise relevant research questions.