Benefits and schemes of prefetching from cloud to fog networks

Abstract

Cloud data services have been in rapidly increasing demand by mobile users, but such services often face the problem of lowered QoS due to the lengthy delay of transmitting large amount of data all the way from the cloud server to the mobile client. In this paper, we propose a prefetching-based approach which utilizes the spare bandwidth and storage space of fog networks to prefetch data files from cloud data servers to edge servers in the fog networks that are much closer to the mobile users, thereby providing better data delivery in terms of transmission delay. We propose six possible prefetching schemes which are designed based on different needs of a priori knowledge about the mobile user's file access patterns. Observing that some prefetching scheme performs better than other schemes when facing a particular type of access pattern, our framework also allows the cloud server and edge servers to dynamically switch to another prefetching scheme if the current prefetching scheme does not fit well with the user's access pattern. We use simulations to evaluate the performance of our proposed prefetching schemes, and the results show that each prefetching scheme can reduce transmission delay, and that the dynamic scheme switching can help improve the performance in terms of prefetching precision and recall ratios.