Improving self-adaptive systems conceptual modeling

Abstract

Self-adaptive Systems (SaSs) operate under uncertainty conditions and have intrinsic properties that have posed some challenges for requirements analysis. Conceptual modeling is useful to requirements analysis because it aids to understand the situation in which a problem occurs. SaSs conceptual modeling is a non-trivial activity because it is necessary to deal with requirements uncertainty, contextual changes, and behavior adaptation. Since conceptual models are built by humans, their quality heavily depends on the humans expertise, which is not a good software engineering practice. Regarding SaSs, the exposure to quality risks increases because of intrinsic characteristic in this class of system. In this paper, we present a SaSs conceptual modeling approach composed of a metamodel and a modeling process. The process defines how to instantiate the metamodel from requirements specifications to create SaSs conceptual models. We performed a controlled experiment with subjects to evaluate our modeling approach effectiveness. As the outcome, we found that our approach had a better performance than an ad hoc approach. The contribution of this paper is a well-defined approach for guiding SaSs conceptual modeling, supported by evidence of its effectiveness by means of an empirical experiment.