Geographic Location-Based Service Reliability Prediction

How to design an effective and efficient reliability prediction method for services is one of the important topics in the research field of Services Computing. With the increasing complexity of network environments, the effect of network environments and other environments related properties on service reliability cannot be neglected any more. However, most of the existing reliability prediction methods focus on the service itself, and have not paid enough attention to the potential impact of external factors on service reliability, which leads to the result that accuracy of predicted reliability of services cannot be guaranteed. To address the problems above, a geographic location-based reliability prediction method (GLBRP) for services is proposed in this paper. Mapping techniques and Pearson Correlation Coefficient are used to classify users based on users´ geographic location information. Deductive reasoning and ontology as well as the calculation of similarity are also employed to clarify services based on services´ location information. Effective feedback can be extracted based on the grouping of users and services. Service reliability is predicted through the smoothing forecasting method of weighted moving series on effective feedback. Simulation results show that the proposed GLBRP method can significantly improve accuracy and efficiency of prediction results compared with other methods.