Prediction of Web Service response time using Grey system model

Abstract

Web Services offered by service provider are expected to maintain a specific level of Quality of Service (QoS) as per the Service Level Agreement (SLA) made between the provider and consumer. The response time is considered as one of the important Quality parameter of a web service. Any breach in SLA in terms of service response time, will lead to a huge penalty in terms of cost and reputation for the service provider. To prevent the penalty caused due to unsatisfactory QoS, it is essential to predict precisely the response time of the web service. Predicting the response time during run time will help in reconfiguring the system, so that the expected performance for the entire system can be preserved. In this paper, we have proposed an agent based Grey system modelling (GM(1,1)) to predict the response time of Web service. The agent based Grey system modelling is capable of predicting the QoS with limited number of past measured QoS values. To enhance the prediction accuracy of grey model GM(1,1), the GM(1,1) model using Fourier series of error residuals has been used.