满足原子事务与QoS端对端约束的服务优化选择方法  

Optimal service selection approach considering both atomic transaction and end-to-end QoS constraints

作　　者：刘海[1] 张卫民[1] 张瞩熹 贾雄[1] 廖蔚[1] 

机构地区：[1]国防科学技术大学计算机学院软件所,湖南长沙410073 [2]总后科研所,北京100071

出　　处：《通信学报》2011年第7期80-92,共13页Journal on Communications

基　　金：国家高技术研究发展计划("863"计划)基金资助项目(2006AA01A123;2010AA012404);国家自然科学基金资助项目(60903042;60736013)~~

摘　　要：提出了一种基于层次服务候选图(SCG)模型优化选择方法,该方法首先利用提出的构造规则和规约规则来构造SCG模型,以建模候选服务之间的事务约束关系,并且该模型关于事务约束的正确性得以证明;其次,提出一种基于广度优先搜索(BFS)的全局QoS感知近优选择算法,该算法基于每个任务的事务属性进行放松剪枝,以控制算法的搜索规模在多项式时间内完成。最后,以网络在线发布的实际Web服务QoS数据和随机产生数据为测试集做模拟实验,结果表明本文提出方法在全局QoS效用和选择成功率方面优于其他相关方法,同时实验通过实现事务约束自动机从实践上也验证了该方法的正确性。An actual SCG（service candidate graph） model-based optimal service selection approach was proposed.Firstly,the approach created transactional constraint relationships among candidates as an actual SCG model with several build-ing rules,whose correctness had been also proved;Then,an optimal QoS-aware service selection algorithm on the basis of BFS（breadth first search） was designed,where a relaxing and pruming method was applied to keep the computation scale in polynomial time.Finally,simulation experiments were conducted with real-world QoS dataset and random data-set,whose results demonstrated the beneficial performance on global QoS utility and outstanding successful selecting ra-tio over other related work.Meanwhile,the correctnesss has also been proved in practice by implementing a transactional automation.

关 键 词：云计算 服务候选图 原子事务 QoS感知服务选择 BFS 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]