Spatiotemporal-Aware Hybrid Prediction Model for Response Time of Web Map Services by Integrating GWR and STARMA

As a non-functional attribute, response time is an important measurement of web service performance. Response time impacts user experience significantly and thus plays an important role in web service selection. However, response time has significantly uncertainty and is hard to predict because it is not only determined by software and hardware performance, but also affected by spatiotemporal distribution of users. In this paper, we take the OGC web map service (WMS) as an example to analyze the correlation between WMS response time and spatiotemporal factors. We collected the monitoring data of response times from thousands of WMSs using a globally distributed monitoring system we developed. Based on data analysis, we propose a hybrid spatiotemporal prediction model that takes the impact of both space and time disparities of user access into account, by integrating geographical weighted regression (GWR) and spatialtemporal auto regressive and moving average (STARMA). Specifically, GWR component simulates macro-level spatiotemporal trends in response time and the STARMA component captures local stochastic variations in spatiotemporal series. By comparing the predicated data with monitoring data from different monitoring sites and at different times, the feasibility was tested. The experimental results show that the proposed model delivers more significant predication accuracy improvement than the classical average model (AVG). The hybrid model also achieves a slight accuracy improvement over GWR and STARMA. We conclude with a discussion of the applicable scenarios of the models.