Temporal Coverage Analysis of Router-Based Cloudlets Using Human Mobility Patterns

Abstract

Responsive applications such as augmented reality require nearby computational offloading units with low latency. The concept of cloudlets is one promising approach that satisfies these requirements. However, due to their wireless range restrictions cloudlets have always been faced with deployment issues of achieving high spatial coverage. In this paper, we look at the coverage issue from an end-user's perspective instead of an established provider perspective: we first argue that temporal coverage of cloudlet accessibility is preferable to spatial coverage for an end- user's experience considering his daily mobility behavior. Next, we investigate what is necessary to achieve a high temporal coverage for an individual user and to what extent is temporal coverage realizable with concepts like router- based cloudlets. To show our hypothesis and understanding the temporal coverage aspect, we collected two comprehensive datasets, an access points dataset with estimated location information and a human mobility dataset consisting of mobility traces from 30 participants within a major city over 4 weeks. Our analysis results show that high temporal coverage can be achieved by a relatively small set of router-based cloudlets since students mainly stay at two places, their homes and university, which represent a large part of the temporal coverage. The remaining rate at which coverage increases heavily depends on the user's mobility patterns. Our findings can be used to place router-based cloudlets at the right locations and estimate the number needed to achieve a certain temporal coverage in urban environments.