An Adaptive Web Services Selection Method Based on the QoS Prediction Mechanism

In recent years, many QoS-based web service selection methods have been proposed. However, as QoS changes dynamically, the atomic services of a composite web service could be replaced with other ones that have better quality. The performance of a composite web service will be decreased if this replacement happens frequently in runtime. Predicting the change of QoS accurately in select phase can effectively reduce this web services “thrash”. In this paper, we propose a web service selection algorithm GFS (Goodness-Fit Selection algorithm) based on QoS prediction mechanism in dynamic environments. We use structural equation to model the QoS measurement of web services. By taking the advantage of the prediction mechanism of structural equation model, we can quantitatively predict the change of quality of service dynamically. Optimal web service is selected based on the predicted results. Simulation results show that in dynamic environments, GFS provides higher selection accuracy than previous selection methods.