A novel energy-efficient and cost-effective task offloading approach for UAV-enabled MEC with LEO enhancement in Internet of Remote Things networks

Abstract

The Internet of Remote Things (IoRT) involves networks of devices deployed in extensive and often remote areas, collecting data for transmission and processing. In such networks, Unmanned Aerial Vehicles (UAVs) gather data, which is then sent to Low Earth Orbit (LEO) satellites for processing. These systems often face significant challenges, particularly in task offloading. Conventional methods typically rely on static routing and scheduling algorithms that do not adapt to changing conditions and usually overlook the complexity of dynamic decision-making in harsh or isolated environments, thus failing to address the critical challenges of energy efficiency and latency. In this paper, we introduce a method comprised of a three-layer architecture. The first layer, the IoRT computing layer, uses Deep Q-Network (DQN) to optimize local decisions based on device constraints and task urgency. The second layer features UAVs serve as Mobile Edge Computing (MEC), which not only processes data but also decides whether to process tasks locally or offload them to LEO satellites, utilizing the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) for this decision-making process. The third LEO Satellite Layer has a high computational capacity to handle offloaded tasks. Simulation results demonstrate notable improvements: compared to another method, the proposed model shows a 14.73 % reduction in energy consumption and a 23.13 % decrease in latency while reducing execution costs by an average of 28.7 %.