摘要
Based on service-oriented architecture（SOA）,a Bellman-dynamic-programming-based approach of service recovery decision-making is proposed to make valid recovery decisions.Both the attribute and the process of services in the controllable distributed information system are analyzed as the preparatory work.Using the idea of service composition as a reference,the approach translates the recovery decision-making into a planning problem regarding artificial intelligence （AI） through two steps.The first is the self-organization based on a logical view of the network,and the second is the definition of evaluation standards.Applying Bellman dynamic programming to solve the planning problem,the approach offers timely emergency response and optimal recovery source selection,meeting multiple QoS （quality of service）requirements.Experimental results demonstrate the rationality and optimality of the approach,and the theoretical analysis of its computational complexity and the comparison with conventional methods exhibit its high efficiency.
基于面向服务的架构(SOA),针对可控分布式信息系统中服务的特点并借鉴服务组合思想,详细分析了信息系统的典型服务过程,提出一种基于贝尔曼动态规划的服务恢复决策算法.通过逻辑层网络自组织和评价标准制定2个步骤,将服务恢复过程中恢复决策问题转化为人工智能领域的规划问题,并对此规划问题使用贝尔曼动态规划算法,解决了确保服务质量(QoS)的前提下,最优服务恢复方案制定与应急响应路径选择问题.仿真实验验证了方法的最优性与合理性,算法复杂性理论分析及与传统方法的性能对比均表明了该方法的高效性.
