Prediction-Based Prefetch Scheduling in Mobile Service Applications

Abstract

The usage of Web or Cloud-based applications on mobile devices suffers from varying link quality, which causes user-perceivable delays and therefore reduces the Quality of Experience (QoE). On the other hand, mobile devices are increasingly feature-rich, allowing to make usage of context data in order to predict network quality based on the user's location. In this paper, we propose a prefetch scheduling algorithm based on network quality predictions, and evaluate it using data collected from real-world field tests. We show that our approach fulfills the expected gain in QoE. Using network quality prediction to optimize data prefetching can improve the userperceived response time by up to 95 %. Our results not only show the feasibility of the proposed algorithm, but also motivate further research in the field of mobility pattern creation, and undermine the importance of location as a part of user context throughout the software stack.