基于验证的自适应系统决策:一种模型驱动的方法

发布时间：2018-04-27 10:39

本文选题：自适应系统 + 优化决策　；参考：《软件学报》2017年07期

【摘要】：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.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.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.
[Abstract]:Adaptive systems need to adjust their behavior according to the runtime context and their own changes. In order to achieve autonomous regulation, adaptive systems must be given the ability to monitor context and self-change at run time, to analyze changes in the degree of requirement satisfaction and to infer to obtain adaptive decision-making. This behavior of online decision-making not only meets the functional requirements, but also ensures that the system meets specific non-functional requirements, such as reliability and performance. An adaptive system optimization decision method based on verification is proposed to meet the non-functional requirements. At the same time, the target model of the system is mapped to the corresponding behavior model, which is represented by the label transfer system, and the reliability of the task is specified by the marking target model, taking the reliability requirement as an example. Then the system behavior model and reliability specification are integrated into discrete time Markov chain with variable state, and candidate adaptive configuration is described as a combination of different variable states. Enables the system to find the optimal decision configuration for a certain type of context. The feasibility and effectiveness of this method are demonstrated by a mobile information system.
【作者单位】：中国科学院数学与系统科学研究院数学研究所;高可信软件技术教育部重点实验室(北京大学);北京大学信息科学技术学院软件研究所;
【基金】：国家重点基础研究发展计划(973)(2015CB352200) 国家自然科学基金(61620106007,91318301)~~
【分类号】：TP311.52

【相似文献】