You Can Enjoy Augmented Reality While Running Around: An Edge-based Mobile AR System

Abstract

Edge computing is proposed to be a promising paradigm to bridge the gap between the stringent computation requirement of realtime mobile augmented reality (MAR) and the constrained computation capacity on our mobile devices. However, prior work on edge-assisted MAR may fail to achieve expected performance in multiple practical cases, e.g., irreparable network disruptions caused by wireless link instability and user-mobility which is a critical characteristic of popular MAR applications. In this paper, we design a new edge-based MAR system named Explorer that enables mobile users to acquire guaranteed MAR offloading performance even under network instability and frequent user-mobility. Additionally, analytical models are developed to provide timely estimation of the sources of object detection staleness. Furthermore, we implement the proposed Explorer in an end-to-end testbed.