Exploiting Mobile Sensor Data for Media Caching in Mobile Edge Networks

Mobile edge network (MEN) is a promising ecosystem within 5G wireless networks. A MEN is often composed of a number of small base stations (sBSs) which can provide cloud computing and storage capabilities in close proximity to users. These new features can be adopted to enhance the Quality of Experience in terms of access delay for multimedia applications on mobile devices, e.g., the smartphones. Content caching is one of the most significant problems in MEN as it can reduce the terminal delay. The existing works often focus on analyzing the network and infrastructure information for the content caching process. However, the mobility and application diversity of mobile users pose significant challenges for content caching in MEN. In this paper, we observe the various sensor data on mobile devices that can be significantly helpful for inferring user preferences on both communicational and computational tasks. Based on this, we propose a novel mobile sensor data based content caching framework for MEN. In the proposed framework, we further design an ambient data collection scheme and an application prediction scheme, which provide the caching nodes capability to predict which applications are about to be launched and what kind of content should be prefetched. We conduct experiments on Android phones and the results show that our proposed framework and methods can achieve accurate content classification and prediction (accuracy >90%).