Prediction-Based Task Offloading for Ultra-Dense LEO Satellite Edge Computing
IEEE Transactions on Mobile Computing · 2026-12 · 25 (12) · 22849-22866
DOI: 10.1109/tmc.2026.3725606 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-12. Verfasst von Gwankyo Nam und 1 weiteren Autor:innen. Die Arbeit nennt 59 Literaturangaben. Gefördert von: Institute of Information & Communications Technology Planning & Evaluation, Next Generation Semantic Communication Network Research Center, National Research Foundation of Korea.
Abstract (Originalsprache)
In this paper, we consider a large-scale low Earth orbit (LEO) satellite edge computing (SEC) system where a ground user generates stochastic tasks and offloads them to satellites, with inter-satellite service migration supported to maintain continuous task processing under satellite mobility. Since the user device and satellites are energy-constrained, energy-efficient offloading and migration are critical design objectives. However, decisions based only on instantaneous conditions can be myopic, triggering frequent service migrations and unnecessary energy consumption. To address this, we propose a predictive task offloading algorithm that leverages the deterministic nature of satellite orbits to forecast satellite availability, channel conditions, and computing capabilities over a multi-slot horizon. To balance the benefit of foresight with the risk of prediction uncertainty, we adopt a $K$ K -slot decision structure that bridges proactive planning and real-time adaptability, and cast each frame-level decision as a shortest-path search on a layered satellite graph capturing time-varying inter-satellite link (ISL) topologies and migration constraints. Theoretical analysis and simulation results validate the effectiveness of the proposed predictive approach and characterize how the prediction window affects performance under uncertainty.
Förderung
- Institute of Information & Communications Technology Planning & Evaluation
- Next Generation Semantic Communication Network Research Center
- National Research Foundation of Korea
- Information Technology Research Center
- Korea Government
Zitieren
APA
Nam, G., & Lee, S. (2026). Prediction-Based Task Offloading for Ultra-Dense LEO Satellite Edge Computing. IEEE Transactions on Mobile Computing, 25(12), 22849-22866. https://doi.org/10.1109/tmc.2026.3725606
Vancouver
Nam G, Lee S. Prediction-Based Task Offloading for Ultra-Dense LEO Satellite Edge Computing. IEEE Transactions on Mobile Computing. 2026;25(12):22849-22866. doi:10.1109/tmc.2026.3725606
BibTeX
@article{Nam2026,
author = {Nam, Gwankyo and Lee, Si-Hyeon},
title = {Prediction-Based Task Offloading for Ultra-Dense LEO Satellite Edge Computing},
journal = {IEEE Transactions on Mobile Computing},
year = {2026},
volume = {25},
number = {12},
pages = {22849--22866},
doi = {10.1109/tmc.2026.3725606}
}
EUVIMED