Research Of QoS-based Cloud Service Composition

With the rapid development of cloud computing technology, since the single cloud service with single function cannot meet the users’various requirements, the technology of cloud service composition gets wide attention. In researches on cloud service composition, few previous studies consider factors such as load balancing, which may impedes the running of cloud service composition in actual operation. As a result, the cloud service composition quality and efficiency may be also affected.Firstly, on the basis of studying a large number of papers about Web service composition, cloud service composition, composition algorithms and other related document, this paper discusses the influence of load balancing and other factors on QoS attributes of cloud service. Combining with the general QoS attributes of service, the QoS model of cloud service is proposed. Meanwhile, considering the influence of cloud service development platform on the QoS attributes’ value, an improved calculation method of QoS attributes of cloud service are proposed. In addition, the calculation model of QoS attributes of cloud service composition which considers the global optimal is proposedSecondly, the genetic algorithm(GA) is applied to get the best solution of cloud service composition which is based on QoS. In order to improve the defect of genetic algorithm, an enhanced initial population generation method is proposed. This method combines a number of chromosomes selected from statistical correlation library and the chromosomes generated with the proportional method into a more optimal initial population. In addition, the selection operation and crossover operation of GA are improved in this paper.Finally, the algorithm is realized and simulating experiments are done in MATLAB. The validity and practicability of QoS-based cloud service composition are proved by the comparison of experimental data.