| 311 | 2 | 113 |
| 下载次数 | 被引频次 | 阅读次数 |
空天地一体化网络(Space-Air-Ground Integrated Network, SAGIN)是未来网络的重要发展范式，利用该网络中的数据训练智能模型将实现多样化的智慧服务。同时，考虑到数据孤岛和隐私安全问题，联邦学习(Federated Learning, FL)的概念被提出，而在资源受限的SAGIN中部署FL算法仍存在训练时间长、学习性能差等挑战。针对这些挑战，围绕SAGIN中非独立同分布(Non-Independent and Identically Distributed, Non-IID)数据场景下的高效FL方法展开研究，提出一种面向SAGIN的3层FL架构，建立设备-无人机、无人机-卫星2级模型聚合机制；针对数据分布异构性问题，进一步揭示设备关联策略对模型误差的影响。基于此，以最小化训练时延与模型误差为目标，结合增益排序机制和凸优化算法联合优化设备关联与带宽分配策略，有效提升FL的训练效率。此外，在卫星服务时间受限的场景下，设计设备动态剔除策略，迭代调整参与训练的设备数量和资源分配结果，确保在有限时间内完成模型训练。仿真结果表明，所提面向SAGIN的FL优化方法能够在保障FL性能的前提下，显著降低训练时延并增加可参与训练的设备数量。
Abstract：Space-Air-Ground Integrated Network(SAGIN) constitute a critical development paradigm for future networks.Training intelligent models using data from this network can enable diverse smart services.However, given the challenges of data silos and privacy concerns, Federated Learning(FL) has emerged as a promising solution.Nevertheless, deploying FL algorithms in resource-constrained SAGIN environments still encounters challenges such as prolonged training times and suboptimal learning performance.This paper investigates efficient FL methods under Non-Independent and Identically Distributed(Non-IID) data scenarios in SAGIN and proposes a three-tier FL architecture tailored for SAGIN.Specifically, it establishes two-level model aggregation mechanisms: one between devices and unmanned aerial vehicles, and another between unmanned aerial vehicles and satellites.Furthermore, it elucidates the influence of device association strategies on model error.Based on these insights, the approach proposed in this paper jointly optimizes device association and bandwidth allocation strategies by leveraging the diminishing marginal gain mechanism and convex optimization techniques, thereby enhancing the training efficiency of FL.Additionally, considering the limited service time of satellites, a dynamic device elimination strategy is designed to iteratively adjust the number of participating devices and resource allocation outcomes, ensuring timely completion of model training within constrained timeframes.Simulation results demonstrate that the proposed method significantly reduces training delay while increasing the number of participating devices and maintaining learning performance.
[1] ZHU G,LIU D,DU Y,et al.Toward an Intelligent Edge:Wireless Communication Meets Machine Learning[J].IEEE Communications Magazine,2020,58(1):19-25.
[2] NIKNAM S,DHILLON H S,REED J H.Federated Learning for Wireless Communications:Motivation,Opportunities,and Challenges[J].IEEE Communications Magazine,2020,58(6):46-51.
[3] QIU C,CHEN Z Y,REN X X,et al.AImers-6G:AI-driven Region-temporal Resource Provisioning for 6G Immersive Services[J].IEEE Wireless Communications,2023,30(3):196-203.
[4] LI H,OTA K,DONG M X.AI in SAGIN:Building Deep Learning Service-oriented Space-Air-Ground Integrated Networks[J].IEEE Network,2023,37(2):154-159.
[5] 李新，王强.6G 网络架构演进及挑战[J].通信与信息技术，2021(4):35-37.
[6] 汪春霆，翟立君，徐晓帆.天地一体化信息网络发展与展望[J].无线电通信技术，2020,46(5):493-504.
[7] LUO L,ZHANG C,YU H F,et al.Energy-efficient Hierarchical Collaborative Learning over LEO Satellite Constellations[J].IEEE Journal on Selected Areas in Communications,2024,42(12):3366-3379.
[8] JIA H Q,WANG Y,WU W.Dynamic Resource Allocation for Remote IoT Data Collection in SAGIN[J].IEEE Internet of Things Journal,2024,11(11):20575-20589.
[9] LIU X L,DENG Y S,MAHMOODI T.Wireless Distributed Learning:A New Hybrid Split and Federated Learning Approach[J].IEEE Transactions on Wireless Communications,2023,22(4):2650-2665.
[10] LI T,SAHU A K,TALWALKAR A,et al.Federated Learning:Challenges,Methods,and Future Directions[J].IEEE Signal Processing Magazine,2020,37(3):50-60.
[11] PARK J,SAMARAKOON S,BENNIS M,et al.Wireless Network Intelligence at the Edge[J].Proceedings of the IEEE,2019,107(11):2204-2239.
[12] MCMAHAN B H,MOORE E,RAMAGE D,et al.Communication-efficient Learning of Deep Networks from Decentralized Data [C]// Proceedings of the 20th International Conference on Artificial Intelligence and Statistics.Ft.Lauderdale:JMLR W&CP,2017:1273-1282.
[13] KAIROUZ P,MCMAHAN H B,AVENT B,et al.Advances and Open Problems in Federated Learning[J].Foundations and Trends? in Machine Learning,2021,14(1-2):1-210.
[14] WANG S Q,TUOR T,SALONIDIS T,et al.Adaptive Federated Learning in Resource Constrained Edge Computing Systems[J].IEEE Journal on Selected Areas in Communications,2019,37(6):1205-1221.
[15] YANG H H,LIU Z Z,QUEK T Q S,et al.Scheduling Policies for Federated Learning in Wireless Networks[J].IEEE Transactions on Communications,2020,68(1):317-333.
[16] LIU L M,ZHANG J,SONG S H,et al.Client-Edge-Cloud Hierarchical Federated Learning[C]// 2020 IEEE International Conference on Communications (ICC).Dublin:IEEE,2020:1-6.
[17] ABAD M S H,OZFATURA E,GUNDUZ D,et al.Hierarchical Federated Learning ACROSS Heterogeneous Cellular Networks [C]// ICASSP 2020-2020 IEEE International Conference on Acoustics,Speech and Signal Processing (ICASSP).Barcelona:IEEE,2020:8866-8870.
[18] FENG J,LIU L,PEI Q Q,et al.Min-Max Cost Optimization for Efficient Hierarchical Federated Learning in Wireless Edge Networks[J].IEEE Transactions on Parallel and Distributed Systems,2022,33(11):2687-2700.
[19] LIM W Y B,NG J S,XIONG Z H,et al.Decentralized Edge Intelligence:A Dynamic Resource Allocation Frame work for Hierarchical Federated Learning[J].IEEE Transactions on Parallel and Distributed Systems,2022,33(3):536-550.
[20] WEN W L,CHEN Z H,YANG H H,et al.Joint Scheduling and Resource Allocation for Hierarchical Federated Edge Learning[J].IEEE Transactions on Wireless Communications,2022,21(8):5857-5872.
[21] ZHANG P Y,ZHANG Y,KUMAR N,et al.Deep Reinforcement Learning Algorithm for Latency-oriented IIoT Resource Orchestration[J].IEEE Internet of Things Journal,2023,10(8):7153-7163.
[22] WANG C,LIU L,JIANG C X,et al.Incorporating Distributed DRL into Storage Resource Optimization of Space-Air-Ground Integrated Wireless Communication Network[J].IEEE Journal of Selected Topics in Signal Processing,2022,16(3):434-446.
[23] HAN D J,HOSSEINALIPOUR S,LOVE D J,et al.Cooperative Federated Learning over Ground-to-Satellite Integrated Networks:Joint Local Computation and Data Offloading[J].IEEE Journal on Selected Areas in Communications,2024,42(5):1080-1096.
[24] ZHAI Z W,WU Q,YU S,et al.FedLEO:An Offloading-Assisted Decentralized Federated Learning Framework for Low Earth Orbit Satellite Networks[J].IEEE Transactions on Mobile Computing,2023,23(5):5260-5279.
[25] HAN D J,FANG W Z,HOSSEINALIPOUR S,et al.Orchestrating Federated Learning in Space-Air-Ground Integrated Networks:Adaptive Data Offloading and Seamless Handover[J].IEEE Journal on Selected Areas in Communications,2024,42(12):3505-3520.
[26] 杨易林.基于联邦强化学习的空天地一体化网络中的流量卸载研究[D].长沙：中南大学，2023.
[27] LIU S L,YU G D,CHEN X F,et al.Joint User Association and Resource Allocation for Wireless Hierarchical Federated Learning with IID and Non-IID Data[J].IEEE Transactions on Wireless Communications,2022,21(10):7852-7866.
[28] KANELLOPOULOS S A,KOUROGIORGAS C I,PANAGOPOULOS A D,et al.Channel Model for Satellite Communication Links Above 10 GHz Based on Weibull Distribution[J].IEEE Communications Letters,2014,18(4):568-571.
[29] 周飞燕，金林鹏，董军.卷积神经网络研究综述[J].计算机学报，2017,40(6):1229-1251.
[30] MILLS J,HU J,MIN G Y.Multi-task Federated Learning for Personalised Deep Neural Networks in Edge Computing[J].IEEE Transactions on Parallel and Distributed Systems,2022,33(3):630-641.
基本信息:
中图分类号:TP18;TN927.2
引用信息:
[1]魏武,徐波,赖海光,等.面向空天地一体化网络的低时延高性能联邦学习方法研究[J].无线电通信技术,2025,51(06):1180-1188.
基金信息:
国家自然科学基金联合基金项目(U2441226)~~
2025-08-20
2025-08-20
2025-08-20