On Evaluating Self-Adaptive Software

Abstract

This article attempts to address the issues of evaluating self-adaptive software systems, an emerging discipline. Since the field is in its early phase and has not produced enough mature systems for consideration, we try to approach the evaluation issue by considering a descriptive model of selfadaptive software based on control systems. Taking inspirations and using the vocabularies from the feedforward and feedback control paradigms, this article will illustrate the analogous properties in self-adaptive software and its evaluation consideration such as stability and robustness. Existing approaches to self-adaptive software take different aspects, ranging from viewing it as new programming paradigm, new architecture style, new modeling paradigm to a new software engineering principle. This article tries to elicit the evaluation consideration from these different aspects.