A model-based framework for self-adaptive software

随着软件系统变得更加相互连接和多样化，架构师预测和设计组件之间的交互的能力越来越弱，将这些问题留在运行时处理。本研究项目的目标是调查、设计和开发一种新的基于模型的框架，以支持自适应软件。该方法可分为三个关键阶段：i)开发方法和工具来定义和描述丰富的语法和语义公共上下文，以允许为反射组件模型和动态组合以及它们之间的交互定义规则；ii)设计一个特征以适应一组可扩展的重构算法；ii)设计一个系统重构引擎来监视系统并决定是否需要改变。我们还相信，所提出的基于模型的机制将执行由重构算法和变更约束组件组成的变更，这些变更约束组件可以独立于驱动变更的应用和适应过程来考虑。最后，我们预计自适应更改引擎将导致决定哪些更改是必要的，以及应该使用什么机制来执行更改。这项研究利用现有软件系统的价值，并提供了工程方法，通过延长公司软件系统的生命周期来增加投资回报。建议的研究将通过工业和开源软件的实验进行验证。我们相信，所开发的技术将帮助行业组织在其进化和维护过程中有效地保持其自适应系统的最新。

As software systems become more interconnected and diverse, architects are less able to anticipate and design interactions among components, leaving such issues to be dealt with at runtime. The objective of this research project is to investigate, design, and develop a novel model-based framework to enable self-adaptive software. The proposed methodology can be divided into three crucial phases: i) developing methodologies and tools to define and describe a rich syntax and semantic common context to allow definition of rules for reflective component models and dynamic composition and interactions between them, ii) designing a feature that accommodates an extensible set of reconfiguration algorithms, and ii) devising a systematic reconfiguration engine that monitors a system and decides whether a change is necessary.We anticipate that such a framework will allow self-adaptive software to be developed with a clean separation of concerns. We also believe that the proposed model-based mechanism will carry out changes which consist of reconfiguration algorithms and change constraint components which can be considered independently from both the application and the adaptation process that drives change. Finally, we anticipate that the adaptive change engine will lead to decisions as to what change is necessary, and what mechanism should be used to carry out that change.This research leverages the value of existing software systems, and provides engineering approaches to increase the return of investment for corporations by prolonging the lifetime of their software systems. The proposed research will be validated through experiments with industrial and open source software. We believe the developed techniques will help industrial organizations to effectively keep their adaptive systems up-to-date during their evolution and maintenance.