UAV-LEO Collaborative Task Offloading in Edge Computing for Internet of Remote Things
Second International Conference on Meta-Networking (MEET 2026), Chofu, Tokyo, Japan, Oct 30, 2026
Summary in three lines
- The study coordinates UAVs and low-Earth-orbit satellites to offload computing for remote IoT devices.
- A multi-agent policy is trained centrally to account for shared satellite capacity and then executed with local observations.
- In simulation, it achieved a better energy–latency trade-off than the reported offloading baselines.
Abstract
The integration of Unmanned Aerial Vehicles (UAVs) and Low Earth Orbit (LEO) satellites in 6G Non-Terrestrial Networks (NTNs) offers a promising paradigm for edge computing of Internet of Remote Things (IoRT). However, dynamic network conditions and shared LEO computing resources make task offloading and resource allocation challenging, particularly due to the coupling among multiple UAVs. In this paper, we investigate the UAV-LEO collaborative task offloading in edge computing for IoRT, which aims to jointly optimize the total energy consumption and task completion delay. To solve this problem, we adopt a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) framework under the Centralized Training with Decentralized Execution (CTDE) paradigm. Specifically, centralized critics capture inter-UAV interactions, while decentralized actors enable efficient local decision making. Simulation results demonstrate that the proposed approach achieves a more favorable trade-off between total energy consumption and task completion delay compared with other baseline schemes.
How to cite
S. Ma, B. Qian, Y. Nan, W. Chi, T. Ma, and M. Tsukada, “UAV-LEO Collaborative Task Offloading in Edge Computing for Internet of Remote Things,” in Second International Conference on Meta-Networking (MEET 2026), Chofu, Tokyo, Japan, 2026.
BibTeX
@inproceedings{Ma2026b,
title = {UAV-LEO Collaborative Task Offloading in Edge Computing for Internet of Remote Things},
author = {Songyuan Ma and Bo Qian and Yuanyu Nan and Weiqi Chi and Ting Ma and Manabu Tsukada},
year = {2026},
date = {2026-10-30},
urldate = {2026-10-30},
booktitle = {Second International Conference on Meta-Networking (MEET 2026)},
address = {Chofu, Tokyo, Japan},
abstract = {The integration of Unmanned Aerial Vehicles (UAVs) and Low Earth Orbit (LEO) satellites in 6G Non-Terrestrial Networks (NTNs) offers a promising paradigm for edge computing of Internet of Remote Things (IoRT). However, dynamic network conditions and shared LEO computing resources make task offloading and resource allocation challenging, particularly due to the coupling among multiple UAVs. In this paper, we investigate the UAV-LEO collaborative task offloading in edge computing for IoRT, which aims to jointly optimize the total energy consumption and task completion delay. To solve this problem, we adopt a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) framework under the Centralized Training with Decentralized Execution (CTDE) paradigm. Specifically, centralized critics capture inter-UAV interactions, while decentralized actors enable efficient local decision making. Simulation results demonstrate that the proposed approach achieves a more favorable trade-off between total energy consumption and task completion delay compared with other baseline schemes.},
keywords = {Other},
pubstate = {published},
tppubtype = {inproceedings}
}