基于验证的自适应系统决策:一种模型驱动的方法

自适应系统需要根据运行时上下文和自身的变化进行其行为的调节.为实现自主调节，自适应系统必须被赋予运行时监测上下文和自身变化，分析需求满足程度的变化，以及推理得到自适应决策的能力.这种在线决策的行为在满足功能需求的同时，还需要保证系统满足特定的非功能需求，如可靠性和性能等.本文提出了一种基于验证的自适应系统优化决策方法，以保证非功能需求的满足.该方法在识别可调节目标以建模自适应机制的同时，将系统的目标模型映射为相应的行为模型，用标签转移系统表示；以可靠性需求为例，用标记目标模型规约任务的可靠性；然后将系统行为模型和可靠性规约整合为带可变状态的离散时间马尔可夫链，将候选自适应配置描述为不同可变状态间的组合；最终通过相关需求的在线验证，使系统找到关于某类上下文的最优决策配置.本文通过一个移动信息系统的案例展示了该方法的可行性和有效性.

Verification Based Decision-Making for Self-Adaptive Systems: A Model-Driven Approach

Self-adaptive systems (SASs) arerequired to be capable of adjusting their behaviors in response to changes in operational contexts and themselves.To implement automatic adjustment, SASs must be endowed by abilities of monitoring changes in contexts and themselves, analyzingchanges of requirements satisfaction and reasoning about adaptation decisions. The behavior of online decision-making needs to assure functional requirements as well as certain non-functional requirements, such as reliability and performance.This paper proposes a verification-basedoptimal decision-making approach for SASs, for assuring the satisfaction of non-functional requirements.The approachmodels adaptation mechanisms by identifying adjustablegoals and mapsgoal models to corresponding behavior models, expressed by Label Transition Systems.It takes reliability requirements as examples and utilizes tagged goal models to specify reliability of tasks. Then, the system behavior modeland reliability specificationsare integrated into Discrete-Time Markov Chains with variablestates, meanwhile adaptation candidates are characterized by combinations of different variable states. Viaonlineverification ofrelated requirements, the system derives the optimal decision of configurations under a certain type of contexts. The feasibility and effectiveness of the approach are illustrated through a mobile information system.