动态网络环境下基于概率的服务组合的可靠性研究

摘要

With the development of Service-Oriented Architecture (SOA), service compositions increasingly appear in various application areas to implement complex tasks. However, in the dynamic network environment, there were influencing factors for the reliability and quality of service compositions. The approaches to service compositions with high reliability in simulated dynamic network environment were presented. A probability model was proposed to quantify the reliability of both serial and parallel service compositions. Three algorithms based on proposed model were designed which included dynamic programming algorithm, improved dynamic programming algorithm and maximum continuity algorithm. These three algorithms were simulated and compared with the random algorithm. The results demonstrated the advantages of proposed algorithms.%随着服务体系架构的不断发展，服务组合正越来越多地出现在各种应用领域以实现复杂的任务。然而，在动态网络环境中，存在着影响服务组合的可靠性和服务质量的各种因素。在模拟动态网络环境的情况下，设计了具有较高可靠性和服务质量的服务组合方法。针对串行和并行的服务组合形式，提出了一个概率模型用以量化服务组合的可靠性，并根据该模型设计了三种算法：动态规划算法、改进的动态规划算法以及连续提供最多服务算法。通过实验模拟了三种算法，并与随机算法进行了比较，结果表明这三种算法具有一定的优势。