开放环境下服务组合的演化行为分析

In the open environment such as the cloud platform and the Internet of things, service composition is regarded as an effective means of fulfilling resource sharing and application inter-connection. In these open environments, however, the construction and execution of a service composition cannot be accomplished at one stroke. It needs to be evolvable at runtime to adapt to unpredictable changes of the requirements and environments. In order to make the self-adaptive evolution of the service composition go smoothly, we need to effectively analyze its evolution behaviors at the runtime stage. Existing approaches to analyzing the behavioral properties of service compositions are mostly based on the control flow models of service compositions. However, the control flow models suffer from the problems of data information missing and probably some unnecessary activity sequence relations. Therefore, these approaches exhibit some deficiencies when they are used to analyze the evolution behaviors of service compositions. To overcome these deficiencies, this project plans to study the approaches to analyzing the evolution behaviors of service compositions based on the program dependence graphs. More specifically, we aim to make some progress in the following aspects: mining the service composition process definition, deciding the compatibility and substitutability of service compositions, and analyzing the migration validity of service composition instances. In addition, we also plan to reify these approaches and techniques in some assistant tools and the corresponding supporting platforms, so as to provide the technical support for the self-adaptive evolution of service compositions.

服务组合被认为是云平台和物联网等开放环境下实现资源共享与应用互联的有效方式。然而，开放环境下服务组合的构造和执行难以一蹴而就，它需要在运行阶段进行自适应的演化以应对需求和环境的非预期变化。为保证服务组合自适应演化的顺利实施，需要在执行阶段对服务组合的演化行为进行有效分析。现有方法多基于控制流模型进行服务组合的行为分析，然而控制流模型存在数据信息缺失以及可能存在不必要的活动顺序关系的问题，这使得相关方法用于服务组合演化行为分析时存在一些不足。为克服这些不足，本项目拟研究基于服务组合程序依赖图的服务组合演化行为分析方法，力图在服务组合过程定义挖掘、服务组合相容性和可替换性判定、服务组合实例迁移有效性分析等方面取得进展与突破，并为这些方法和技术配备相应辅助工具和支撑平台以提供对服务组合自适应演化的技术支持。

提出了一套以“服务组合程序依赖图”为核心的服务组合演化时行为性质分析方法，主要包括基于活动间依赖关系的服务组合过程模型挖掘，基于活动约束关系的服务组合行为一致性分析度量，基于动态依赖图的服务组合实例迁移有效性分析等技术。依据这些技术，开发了一系列符合业界标准的软件支撑工具，为下一代服务组合演化支撑系统的研制提供了技术储备。