Improved Adaptation of Web Service Composition Based on Change Impact Probability

Abstract

Due to the dynamic nature of environment in service computing, it becomes more important to make the Web service composition able to self-adapt to changes in its environment. Achieving this goal is a challenging task, as the performance of a composite Web service will be decreased if the adaptation happens frequently in runtime. In this paper, we improve adaptation of Web service composition by predicting the probability that a change will actual affect the running composite service, which is named change impact probability (CIP), and provide a methodology based on a comprehensive QoS model of a QoS change and an execution state model of an executing composite service for computing the CIP. To bring the proposed approach to fruition, we develop a prototype system and apply the approach to a loan rate query service for illustration.