SatFed: A Resource-Efficient LEO-Satellite-Assisted Heterogeneous Federated Learning Framework

Engineering ›› 2025, Vol. 54 ›› Issue (11) : 115 -126. DOI: 10.1016/j.eng.2025.07.020

Research

Article

Yuxin Zhang ^a^,^b
, Zheng Lin ^a^,^c
, Zhe Chen ^a
, Zihan Fang ^a^,^d
, Xianhao Chen ^c
, Wenjun Zhu ^a^,^b
, Jin Zhao ^a^,^b
, Yue Gao ^a^,^*

Author information +

History +

Received	Accepted	Published
2024-09-20	2025-07-14	2025-07-24
Issue Date	Revised Date
2025-07-24	2025-06-17

PDF (2686KB)

Abstract

Traditional federated learning (FL) frameworks rely heavily on terrestrial networks, whose coverage limitations and increasing bandwidth congestion significantly hinder model convergence. Fortunately, the advancement of low-Earth-orbit (LEO) satellite networks offers promising new communication avenues to augment traditional terrestrial FL. Despite this potential, the limited satellite–ground communication bandwidth and the heterogeneous operating environments of ground devices—including variations in data, bandwidth, and computing power—pose substantial challenges for effective and robust satellite-assisted FL. To address these challenges, we propose SatFed, a resource-efficient satellite-assisted heterogeneous FL framework. SatFed implements freshness-based model-prioritization queues to optimize the use of highly constrained satellite–ground bandwidth, ensuring the transmission of the most critical models. Additionally, a multigraph is constructed to capture the real-time heterogeneous relationships between devices, including data distribution, terrestrial bandwidth, and computing capability. This multigraph enables SatFed to aggregate satellite-transmitted models into peer guidance, improving local training in heterogeneous environments. Extensive experiments with real-world LEO satellite networks demonstrate that SatFed achieves superior performance and robustness compared with state-of-the-art benchmarks.

Keywords

Low-Earth-orbit satellite networks / Distributed machine learning / Federated learning / System heterogeneity

Cite this article

Download citation ▾

Yuxin Zhang, Zheng Lin, Zhe Chen, Zihan Fang, Xianhao Chen, Wenjun Zhu, Jin Zhao, Yue Gao. SatFed: A Resource-Efficient LEO-Satellite-Assisted Heterogeneous Federated Learning Framework. Engineering, 2025, 54(11): 115-126 DOI:10.1016/j.eng.2025.07.020

登录浏览全文

4963

注册一个新账户忘记密码

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	Liu J, Ren J, Zhang Y, Yue S, Zhang Y.SESAME: a resource expansion and sharing scheme for multiple edge services providers.IEEE/ACM Trans Netw 2024; 32(4):3189-3204.

[2]

Zhang X, Xu H, Zou L, Duan J, Wu C, Xue Y, et al.Rosevin: employing resource-and rate-adaptive edge super-resolution for video streaming.In: Proceedings of the IEEEAnnual Joint Conference: INFOCO M, IEEE Computer and Communications Societies; 2024 May 20–23; New York City, N Y, USA. New York City: IEE E; 2024. p. 1–10.

[3]	Ren J, Liu J, Zhang Y, Li Z, Lyu F, Wang Z, et al.An efficient two-layer task offloading scheme for MEC system with multiple services providers.In: Proceedings of IEEEConference on Computer Communications; 2022 May 2–5; online. New York City: IEE E; 2022. p. 1519–28.

[4]

Zheng T, Li A, Chen Z, Wang H, Luo J.AutoFed: heterogeneity-aware federated multimodal learning for robust autonomous driving.In: Proceedings of the 29th Annual International Conference on Mobile Computing and Networking; 2023 Oct 2–6; Madrid, Spain. New York City: Association for Computing Machinery (ACM); 2023. p. 1–15.

[5]	Marelli L, Testa G.Scrutinizing the EU general data protection regulation.Science 2018; 360(6388):496-498.

[6]	Rodio A, Faticanti F, Marfoq O, Neglia G, Leonardi E.Federated learning under heterogeneous and correlated client availability.IEEE/ACM Trans Netw 2023; 32(2):1451-1460.

[7]

McMahan B, Moore E, Ramage D, Hampson S, Arcas BA.Communication-efficient learning of deep networks from decentralized data.In: Proceedings of the 20th International Conference on Artificial Intelligence and Statistics; 2017 Apr 20–22; Ft. Lauderdale, F L, USA. New York City: MLResearch Press; 2017. p.1273–82.

[8]	Sacco A, Flocco M, Esposito F, Marchetto G.Owl: congestion control with partially invisible networks via reinforcement learning.In: Proceedings of IEEEConference on Computer Communications; 2021 May 10–13; online. New York City: IEE E; 2021. p. 1–10.

[9]	Öhl Pén, Skubic B, Rostami A, Fiorani M, Monti P, Ghebreten-sa Zé, et al.Data plane and control architectures for 5G transport networks.J Lit Technol 2016; 34(6):1501-1508.

[10]	Yang D, Zhang W, Ye Q, Zhang C, Zhang N, Huang C, et al.DetFed: dynamic resource scheduling for deterministic federated learning over time-sensitive networks.IEEE Trans Mobile Comput 2024; 23(5):5162-5178.

[11]	Xu J, Kishk MA, Alouini MS.Space–air–ground–sea integrated networks: modeling and coverage analysis.IEEE Trans Wirel Commun 2023; 22(9):6298-6313.

[12]	Zhai Z, Wu Q, Yu S, Li R, Zhang F, Chen X.FedLEO: an offloading-assisted decentralized federated learning framework for low Earth orbit satellite networks.IEEE Trans Mobile Comput 2024; 23(5):5260-5279.

[13]	Zhang Y, Wu Q, Lai Z, Li H.Enabling low-latency-capable satellite–ground topology for emerging LEO satellite networks.In: Proceedings of IEEEConference on Computer Communications; 2022 May 2–5; New York City, N Y, USA. New York City: IEE E; 2022. p. 1329–38.

[14]	Ahmmed T, Alidadi A, Zhang Z, Chaudhry AU, Yanikomeroglu H.The digital divide in Canada and the role of LEO satellites in bridging the gap.IEEE Commun Mag 2022; 60(6):24-30.

[15]	Lai Z, Li H, Wang Y, Wu Q, Deng Y, Liu J, et al.Achieving resilient and performance-guaranteed routing in space–terrestrial integrated networks.In: Proceedings of IEEEConference on Computer Communications; 2023 May 17–20; New York City, N Y, USA. New York City: IEE E; 2023. p. 1–10.

[16]	Apollonio P, Caini C, Lülf M.DTN LEO satellite communications through ground stations and GEO relays.R. Dhaou, A.L. Beylot, M.J. Montpetit, D. Lucani, L. Mucchi (Eds.), Personal satellite services, Springer, Cham 2013; 1-12.

[17]	i Perf—the ultimate speed test tool for TCP, UD Pand SCTP [Internet].Reims: iKoula; [cited 2025 Jun 18]. Available from: https://iperf.fr/.

[18]	He K, Zhang X, Ren S, Sun J.Deep residual learning for image recognition.In: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition; 2016 Jun 27–30; Las Vegas, N V, USA. New York City: IEE E; 2016. p. 70–8.

[19]	Li T, Sahu AK, Zaheer M, Sanjabi M, Talwalkar A, Smith V.Federated optimization in heterogeneous networks.In: Proceedings of Machine Learning and Systems 2; 2020 Mar 2–4; Austin, T X, USA. Indio: MLSys Proceedings; 2020. p. 429–50.

[20]	Chen S, Li B.Towards optimal multi-modal federated learning on non-IID data with hierarchical gradient blending.In: Proceedings of IEEEConference on Computer Communications; 2022 May 2–5; London, U K. New York City: IEE E; 2022. p. 1469–78.

[21]	Tan AZ, Yu H, Cui L, Yang Q.Towards personalized federated learning.IEEE Trans Neural Netw Learn Syst 2022; 34(12):9587-9603.

[22]	Krizhevsky A, Hinton G.Learning multiple layers of features from tiny images. University of Toronto, Toronto (2009)

[23]	Liao Y, Xu Y, Xu H, Wang L, Qian C.Adaptive configuration for heterogeneous participants in decentralized federated learning.In: Proceedings of IEEEConference on Computer Communications; 2023 May 17–20; Hoboken, N J, USA. New York City: IEE E; 2023. p. 1–10.

[24]	Diao E, Ding J, Tarokh V.HeteroFL: computation and communication efficient federated learning for heterogeneous clients.2020. arXiv: 2010.01264.

[25]	Basford PJ, Johnston SJ, Perkins CS, Garnock-Jones T, Tso FP, Pezaros D, et al.Performance analysis of single board computer clusters.Future Gener Comput Syst 2020; 102:278-291.

[26]	Li T, Hu S, Beirami A, Smith V.Ditto: fair and robust federated learning through personalization.In: Proceedings of the 38th International Conference on Machine Learning; 2021 Jul 18–24; online. New York City: MLResearch Press; 2017. p.6357–68.

[27]	Dinh TC, Tran NH, Nguyen TD.Personalized federated learning with moreau envelopes.In: Proceedings of the 34th International Conference on Neural Information Processing Systems; 2020 Dec 6–12; Vancouver, B C, Canada. Red Hook: Curran Associates Inc.; 2020. p. 21394–405.

[28]	Ye R, Ni Z, Wu F, Chen S, Wang Y.Personalized federated learning with inferred collaboration graphs.In: Proceedings of 40th International Conference on Machine Learning; 2023 Jul 23–29; Honolulu, H I, USA. New York City: MLResearch Press; 2023. p. 39801–17.

[29]	Xiao H, Rasul K, Vollgraf R.Fashion-MNIST: a novel image dataset for benchmarking machine learning algorithms.2017. arXiv: 1708.07747.

[30]	Kingma DP, Ba J.Adam: a method for stochastic optimization.2014. arXiv: 1412.6980.

[31]	Xie C, Koyejo S, Gupta I.Asynchronous federated optimization.2019. arXiv: 1903.03934.

[32]	Smith V, Chiang CK, Sanjabi M, Talwalkar AS.Federated multi-task learning.In: Proceedings of the 31st International Conference on Neural Information Processing Systems; 2014 Dec 4–9; Long Beach, C A, USA. Red Hook: Curran Associates Inc.; 2014. p. 4427–37.

[33]	Nguyen J, Malik K, Zhan H, Yousefpour A, Rabbat M, Malek M, et al.Federated learning with buffered asynchronous aggregation.In: Proceedings of The 25th International Conference on Artificial Intelligence and Statistics; 2022 Mar 28–30; online. New York City: MLResearch Press; 2022. p.3581–607.

[34]	Wang Z, Xu H, Liu J, Huang H, Qiao C, Zhao Y.Resource-efficient federated learning with hierarchical aggregation in edge computing.In: Proceedings of the IEEEAnnual Joint Conference: INFOCO M, IEEE Computer and Communications Societies; 2021 May 10–13; online. New York City: IEE E; 2021. p. 1–10.

[35]	Sun Y, Peng M, Zhang S, Lin G, Zhang P.Integrated satellite–terrestrial networks: architectures, key techniques, and experimental progress.IEEE Netw 2022; 36(6):191-198.

[36]	Cui G, Duan P, Xu L, Wang W.Latency optimization for hybrid GEO–LEO satellite-assisted IoT networks.IEEE Internet Things J 2023; 10(7):6286-6297.

[37]	Ding C, Wang JB, Cheng M, Lin M, Cheng J.Dynamic transmission and computation resource optimization for dense LEO satellite assisted mobile-edge computing.IEEE Trans Commun 2023; 71(5):3087-3102.