Modeling mobility and content similarity for efficient device-to-device data sharing -

Abstract

Creation and availability of user-generated content is facilitated by the widespread use of smartphones, which provide mobility and connectivity along with user friendly graphical interfaces. Considering that the majority of internet traffic is produced by end users, network offloading using device-to-device (D2D) communications is an attractive approach to enhance network performance. For any two users to share data via D2D links, they need to be in close proximity for a long enough period of time, and have similar content interests. In this work, we model mobility and content similarity between smartphones using an experimental study focused on wireless D2D content sharing applications. In our study, engineering students from the American University of Beirut filled a users content interest survey and downloaded an Android application on their smartphones in order to collect location and neighbor discovery data. The survey responses are used to perform content similarity analysis, while the data collected by the Android application is used to empirically model and analyze the number of contacts and inter-contacts between devices, and the contact and inter-contact durations, as well as to identify hub locations. The obtained models and insights can serve as key inputs to generate more accurate performance results for wireless network designs with D2D communications.