支持服务关联的全局最优服务选择算法

摘要

Aiming at the problem that a service-oriented military electronic information system' s QoS of a candidate service relies on other candidate services,a method for global optimization of service selection with the consideration of service correlation is proposed.The method transforms the problem of dynamic web service selection with QoS global optimal into a multiobjective services composition optimization with QoS constraints,and the multiobjective estimation of distribution algorithm based on independent component analysis (ICA) is utilized to produce a set of optimal Pareto services composition with constraint principle by means of optimizing various objective functions simultaneously.The results of the compared experiment indicate that the multiobjective estimation of distribution algorithm based on ICA can describe the correlation between candidate services through the probabilistic graphical model,and the algorithm has the better performance in solving the service selection problem with service correlation,which confirms its feasibility and efficiency.%针对面向服务的军事综合电子信息系统候选服务的服务质量往往依赖于其他候选服务的问题,提出了一种支持服务关联的全局最优服务选择方法.该方法将基于QoS的军事信息服务选择问题建模为带QoS约束的多目标组合优化问题,采用基于独立成分分析(ICA)的多目标分布估计算法同时优化多个目标函数,最终产生一组满足约束条件的Pareto最优解服务组合集.对比实验结果表明,基于ICA的多目标分布估计算法通过概率图模型可以描述服务间的相互关系,可以更好地解决存在服务关联的组合服务选择问题,具有良好的有效性和可行性.