基于FICAS的积极调停性能优化  

Performance Optimization on Active Mediation for FICAS

作　　者：李丹[1] 姚国祥[1] 车国海[1] 

机构地区：[1]广东广州暨南大学网络中心,510632

出　　处：《微计算机信息》2006年第02X期239-243,共5页Control & Automation

基　　金：广东省自然科学基金重点项目(010421)

摘　　要：基于流的服务合成架构——FICAS,简单描述服务合成中积极调停的动态加载移动类的基本思想,重点介绍了使积极调停性能优化的DavidLiu的LDS(最大数据量)算法,分析了该算法确定移动类最优加载位置的优点和局限性。针对LDS算法的局限性,提出了更具通用性和实用性的NN_LDS(神经网络最大数据量)算法,解决了任意移动类在输入数据大小已知,输出数据大小未知条件下,最优加载位置的确定问题,从而实现了积极调停性能优化。An overview of the basic ideas underlying active mediation technology and mobile classes dynamically loaded by active mediation for service composition infrastructure based on flow is given. David Liu＇s LDS algorithm used to optimize active mediation performance is emphasized and is also analyzed for its merits and limits in dynamically loading mobile classes. Facing LDS＇s limits, a more general and practical NNLDS algorithm is proposed here, which can locate mobile classes at optimal placement underlying input data sizes known and output data sizes unknown for them. Thus,active mediation performance is further optimized .

关 键 词：FICAS 积极调停 移动类 最大数据量算法 神经网络最大数据量算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]