Evaluation of the State-Constraint Transition Modelling Language: A Goal Question Metric Approach

Self-adaptive systems (SAS) are exceptional systems, on account of their versatile composition, dynamic behavior and evolutive nature. Existing formal languages for the specification of SAS focus on adapting system elements to achieve a target goal, following specific rules, without much attention on the adaptation of requirements themselves. The State-Constraint Transition (SCT) modeling language enables the specification of dynamic requirements, both at the domain and application level, as a result of space or time variability. This language, evaluated in this paper, enables the specification of a variety of requirement types, for SASs from different domains, while generating a configuration, all configurations, and number of possible configurations, in milliseconds. This paper presents these results, namely; expressiveness, domain independence and scalability, from the viewpoint of designers and domain engineers, following a goal-question-metric approach. However, being primarily based on constraint programming (CP), the language suffers from drawbacks inherited from this paradigm, specifically time related requirements, like (e.g. order, frequency and staged requirements).