Performance Evaluation of Edge Computing-Aided IoT Augmented Reality Systems
Weiyang Qian, Rodolfo W. L. Coutinho · 2022
Envisioned mobile augmented reality (MAR) ushers a new plethora of smart applications. However, the resource-constrained nature of head-mounted devices (HMDs) has limited the development of MAR systems. In this regard, edge computing has emerged as a promising solution for the processing of MAR computer-intensive tasks. In edge-aided MAR systems, HMDs will offload to edge nodes part of the acquired video frames, which reduces the latency and energy cost for video analytics in MAR systems. In this paper, we devise a queuing theory-based mathematical framework for guiding the design of MAR systems. The proposed mathematical framework models the characteristics of HMDs and edge devices, and the different network conditions. It serves as a tool for directing the decision-making in the design of MAR systems under different conditions and applications. Extensive numerical evaluations show that offloading frames to edge servers at a proper rate can significantly reduce the total average latency while a higher offloading rate incurs lower energy costs at MAR devices.