ProbaSAS: Modeling and Decision-Making Approach for Self-Adaptive Software Systems under Uncertainty

Abstract

Self-adaptive software (SAS) systems are gaining increasing popularity in recent years. However, the changing and dynamic running environment and the diverse user requirements have introduced uncertainty, especially the random uncertainty into behaviors of the SAS systems. And the above uncertainty has posed huge challenges in software modeling and decision-making of the SAS systems. For this end, this paper presents ProbaSAS: an MDP (Markov Decision Process) based approach for SAS modeling and decision making under uncertainty. Firstly, the uncertainty within the SAS systems has been systematically analyzed, and the modeling and decision-making framework is proposed. Then, the modeling approach for three kinds of uncertainty (i.e., the probabilistic behaviors, the non-deterministic processes, and the non-functional characteristics) is created based on the MDP model. Finally, the self-adaptation reasoning and decision-making approach for two kinds of self-adaptation (i.e., the structure self-adaptation and the behavior self-adaptation) is proposed based on the probabilistic model checking technique. Taking the Ship-Supplying Information System as an example, we have evaluated the effectiveness of the ProbaSAS approach in uncertainty modeling and decision-making of the SAS systems.