Empirical evaluation of tropos4AS modelling

Our work addresses the challenges arising in the development of self-adaptive software, which has to work autonomously in an unpredictable environment, fulfilling the objectives of its stakeholders, while avoiding failure. In this context we developed the Tropos4AS framework, which extends the AOSE methodology Tropos to capture and detail at design time the specific decision criteria needed for a system to guide selfadaptation at run-time, and to preserve the concepts of agent and goal model explicitly along the whole development process until run-time. In this paper, we present the design of an empirical study for the evaluation of Tropos4AS, with the aim of assessing the modeling effort, expressiveness and comprehensibility of Tropos4AS models. This experiment design can be reused for the evaluation of other modeling languages extensions.