一种基于动态Skyline和遗传粒子群优化的云服务组合方法

Approach for Cloud Service Composition Based on Dynamic Skyline and Genetic Particle Swarm Optimization

提出一种改进的基于动态Skyline和多种群遗传粒子群优化的云服务组合优化方法,旨在解决动态、不确定环境下大规模云服务组合优化问题.对云服务组合和服务质量(QoS)形式化描述,提出一种云服务组合优化模型;对Skyline操作进行建模的基础上,设计Skyline云服务动态更新算法,以满足云服务因临时加入、退出及QoS变化而引起的动态性和不确定性需求;最后,设计一种新的云服务组合优化算法,算法采用动态Skyline操作和用户约束降低问题求解空间,并基于种群相似性和遗传操作进行防早熟收敛处理.通过真实数据集和随机数据集的大量仿真实验,结果验证了本文算法的可行性和有效性.

A new improved cloud service composition （ CSC ） optimization method was proposed, which was based on the dynamic skyline and muff-colony genetic particle swarm optimization, to solve the large-scale CSC optimization problem with dynamic and un- certain environments. Firstly, on the basis of the formalization description of CSC and QoS, a cloud service composition optimal model was proposed. Secondly, a skyline dynamic updating algorithm based on the Skyline operation modeling was designed which meets the dynamic and uncertain requirements caused by the temporary join, exit and QoS change of cloud service. Finally, a novel cloud service composition optimal algorithm was proposed, the solution space was reduced through the dynamic Skyline operation and the user con- straint, and the problem of premature convergence was solved by using the population similarity and genetic operation. A large number of simulation experiments were carded on the actual and random data set, and its experiment results validate the feasibility and effi- ciency of the algorithms.