自适应软件系统的无缝演化与环境感知技术研究

像因特网和物联网这样的开放计算环境下的软件系统常常面临着环境和用户需求的变化。其中许多系统需要能够在不中断服务的前提下自动地适应这些变化。既有的软件构造和运行支撑技术缺乏对自适应的系统支持。为此本项目拟研究（1）较为通用、灵活的自适应软件系统体系结构，不但支持反应式（Reactive）的自适应，也支持前瞻式（Proactive）的自适应；（2）软件系统无缝在线演化技术，即在保证系统一致性的前提下，更好地控制在线演化对系统正常服务造成的干扰，并使其更加及时；（3）面向环境感知的软件方法，以先验环境建模、应用情境规约和环境感知中间件来系统地支持准确高效的环境感知。本项目还将为这些技术开发相应的支撑工具。在此基础上，逐步将自适应支持引入到主流的软件开发和运行支撑技术体系中去，为研制下一代支持软件自适应的中间件系统提供技术准备。

Software systems working in open environments such as the Internet and Cyber-Physical Systems are always facing changes in their underlying environment and in user requirements. Many of these systems have to automatically adapt themselves to these changes at runtime. This research is aimed at facilitating the engineering of self-adaptive software systems. Expected outcomes include (1) a general and flexible architecture for self-adaptive software systems, which supports proactive adaptations in addition to reactive ones, (2) an approach to seamless software evolution, i.e., updating a part of a software system at runtime in a timely and low-disruptive way and without any compromise in system consistency, and (3) a disciplined method for software environment perception, which leverages environmental models a priori to recognize systems' dynamic contextual situation in an accurate and efficient way. Tools enabling the application of these techniques will also be developed. As our long-term goal is to invent next-generation middleware systems with comprehensive support for self-adaptation, techniques' applicability to the mainstream software development will be emphasized in the research.