一种基于业务生成图的Web服务工作流构造方法  

An Approach to Constructing Web Service Workflow Based on Business Spanning Graph

作　　者：胡春华[1] 吴敏[1] 刘国平[1] 徐德智[1] 

机构地区：[1]中南大学信息科学与工程学院

出　　处：《软件学报》2007年第8期1870-1882,共13页Journal of Software

基　　金：the National Natural Science Foundation of Chinaunder Grant No.60674016(国家自然科学基金);the National High-Tech Research and Development Plan of Chinaunder Grant No.2006AA04Z172(国家高技术研究发展计划(863));the National Science Fund for Distinguished Young Scholars of Chinaunder Grant No.60425310(国家杰出青年科学基金);the Natural Science Foundation of Hu’nan Province of Chinaunder Grant No.05JJ40118(湖南省自然科学基金)

摘　　要：针对互联网中Web服务具有动态变化且迅速增长的特点,提出了一种面向用户需求的服务工作流构造模型.该模型将功能相同或相似的服务聚集成一类服务集合,每类服务集合采用生成树的方式组织,并依据工作流的业务逻辑关系形成业务生成图;同时,在重定义粒子群算法的位置、速度、加/减法和乘法的基础上,结合遗传算法中的交叉、变异操作,设计了基于混合粒子群的QoS(quality of service)调度方法,保证在可选服务不断增长时能够满足用户的个性化需求.实验结果表明,该模型能够有效地屏蔽组成工作流的Web服务物理上的变化与差异,较好地组合了Internet中的Web服务资源,适合于虚拟计算环境的应用要求.Based on the fact that Web service dynamically changes and rapidly increases in the Internet, a user-oriented service workflow constructing model is proposed. The same or similar function services are accumulated into a kind of service set in this model, which is organized by the spanning tree, and the business spanning graph is formed according to workflow＇s business logic relation. At the same time, on the basis of redefining the position, velocity, addition, subtraction and multiplication of particle swarm algorithm, combining with the cross and mutation operations in genetic algorithm, the QoS （quality of service）scheduling algorithm based on hybrid particle swarm optimization algorithm （HPSOA） is designed to satisfy different multi-QoS demands when Web service rapidly increases. Experimental results demonstrate that the constructing model could effectively shield the physical varieties and differences of Web service, and preferably combine the Web service resource in the Internet. It is suitable for the application demands in the virtual computing environment.

关 键 词：WEB服务 工作流 业务生成图 QoS(quality of service)调度 混合粒子群算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]