Dynamic Energy-Efficient Computation Offloading in NOMA-Enabled Air-Ground Integrated Edge Computing

• Published in: IEEE internet of things journal p. 1
• Main Authors: Li, Heng, Chen, Ying, Li, Kaixin, Yang, Yaozong, Huang, Jiwei
• Format: Journal Article
• Language: English
• Published: IEEE 16.08.2024
• Subjects: Atmospheric modeling; Autonomous aerial vehicles; Computational modeling; Dynamic Task Offloading; Internet of Things; Internet of Things (IoT); Mobile Edge Computing (MEC); NOMA; Non-orthogonal Multiple Access (NOMA); Servers; Task analysis; Unmanned Aerial Vehicle (UAV)
• ISSN: 2327-4662
• DOI: 10.1109/JIOT.2024.3438772

With the swift progress of Internet of Things (IoT) technologies, the number of IoT devices has grown exponentially, leading to an increasing demand for computational power and system stability. Mobile Edge Computing (MEC) is a powerful solution that allows IoT devices to offload data to the edge for computing. In situations involving disasters or complex terrains, establishing ground-based stations may be challenging in providing computational services. Edge computing frameworks built with unmanned aerial vehicles (UAVs) and high-altitude platforms (HAPs) can provide airborne computational services for IoT devices situated in environments with disasters or complex terrains. In this paper, we design a three-tier framework consisting of ground users (GUs), UAVs, and HAP, offering MEC services for GUs. Considering the randomness and dynamism of task arrivals and the wireless communication quality of devices, we propose an algorithm supporting Non-Orthogonal Multiple Access (NOMA) communication in aerial access networks. The objective of the algorithm is to reduce the overall energy consumption of the system while ensuring system stability. Employing stochastic optimization techniques, we convert the task offloading and resource allocation problem into several parallel solvable sub-problems. We also provide a theoretical analysis of the algorithm. Through a series of comparative experiments, we demonstrate the feasibility and effectiveness of our proposed Dynamic Energy-Efficient Computation Offloading (DEECO) algorithm.