QoS-Aware Dynamic Service Composition

The Web service technology is employed by more and more organizations and research institutions to realize data sharing and value-added,and the Web services deployed in an open environment continue to emerge.Given that the limitation of a single service,it is difficult to adapt in diverse business scenarios,so these services need to be combined to achieve the complex function.Most of the existing research works about the Web service composition on the premise that the quality of Web service is reliable,and the QoS(Quality of Service)provided by service providers are real.In fact,the QoS information of a Web service may be constantly changed due to the network load and the influence of geographical location.In order to measure the effects of these changes on the QoS of Web service,the QoS must be monitored and collected in real time.In addition,some of the research works on service composition are mostly based on the least cost and the shortest execution time,while the global QoS constraint is ignored.Otherwise,the quality attributes that characterize the service delivery capabilities are often of concern to users,and these attributes should not be ignored in Web service composition systems.With the increasing attention to the efficiency of the combination,users also put forward higher requirements on the search performance of the combined algorithm,and the service composition process must be accomplished in a reasonable time.This paper focuses on how to ensure the effectiveness of QoS based service composition system in dynamic environment and how to improve the efficiency of service composition.From the aspects of monitoring and updating QoS information,modeling and analyzing the process of service composition,the main research works includes the following aspects.(1)A QoS information monitoring system based on multi-agent is proposed according to the complex characteristics of service execution environment.The dynamic characteristics of QoS information are considered during the running process.A multi-agent cooperation model based on mental calculation and contract net protocol and its implementation are given to dynamically collect and update QoS information during Web service operation process.(2)A local service filtering strategy based on AHP(Analytic Hierarchy Process)method is proposed in this paper,and considering the characteristics of service composition,the analysis model is established to sort local services based on the QoS of the service,so as to choose the former Top n services in each type of atom service sets.(3)In order to ensure the combination of services to meet the global optimal QoS,the culture genetic algorithm is put forward to deal with the problem of service composition.The genetic algorithm is regarded as the social population space of culture algorithm,and the diversity and similarity of chromosomes exclusion strategy is used to expand the diversity of the population and accelerate the convergence speed,which can ensure the efficiency of service composition system.Finally,experiments are carried out to verify the efficiency of the multi Agent cooperation based on mind computing and contract net protocol,and the effect of local service selection on global service optimization is verified and analyzed.Moreover,the search efficiency and execution time of the improved culture genetic algorithm are analyzed,and the experimental results show that the proposed method is very feasible.