期刊
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY
卷 69, 期 11, 页码 13631-13645出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TVT.2020.3020982
关键词
Object detection; Servers; Energy consumption; Optimization; Quality of service; Image resolution; Image edge detection; Augmented reality; energy efficiency; multi-access edge computing; QoS guaranteed; object detection; optimization
资金
- Institute of Information & Communications Technology Planning & Evaluation (IITP) - Korea Government (MSIT) [2018-0-00691]
- IT R&D program of MSIT (Ministry of Science and ICT) [2018-0-00226]
- Singapore Ministry of Education (MOE) Tier 1 [RG16/20]
In this study, we focus on improving the energy efficiency of multiple mobile augmented reality (MAR) devices (MDs) with different MAR applications, which are connected to a single multi-access edge computing (MEC) server, through the centralized orchestration of the MEC server. To achieve this, a trade-off between the accuracy, latency, and energy consumption of each MD is considered as intertwined costs. Accordingly, we minimize a sum of intertwined costs including the energy consumption, latency and accuracy loss of multiple MDs, while satisfying the maximum latency constraint and the minimum accuracy constraint of each MAR application. With rigorous analysis, we design a theoretical framework for the optimization of MD performance by jointly managing the frame resolution of the image to offload to the MEC server from MDs, and the computation capacity distribution over MDs in the case of the limited computation capacity of the MEC server. Our numerical analysis validates that the proposed scheme significantly reduces the number of service-level agreement violations, such as exceeding the maximum latency constraint, compared with the existed algorithm. Specifically, the proposed work significantly improve costs of service latency, accuracy loss, and energy consumption of MDs for MEC-assisted MAR services compared with baseline schemes.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据