From means-end analysis to proactive means-end reasoning

Abstract

Self-adaptation is a prominent property for developing complex distributed software systems. Notable approaches to deal with self-adaptation are the runtime goal model artifacts. Goals are generally invariant along the system lifecycle but contain points of variability for allowing the system to decide among many alternative behaviors. This work investigates how it is possible to provide goal models at run-time that do not contain tasks, i.e. The description of how to address goals, thus breaking the design-time tie up between Tasks and Goals, generally outcome of a means-end analysis. In this vision the system is up to decide how to combine its available Capabilities: the Proactive Means-End Reasoning. The impact of this research line is to implement a goal-oriented form of self-adaptation where goal models can be injected at runtime. The paper also introduces MUSA, a Middleware for User-driven Service self-Adaptation.