With the development of low-altitude applications, there is a surge in demand for reliable unmanned aerial vehicle (UAV) communications with high data rate. Meanwhile, the beamforming technique can significantly improve the ground-to-air transmission quality through the phase and amplitude adjustment with the antenna array. Due to the high mobility of the UAV, it is vital to guarantee that the directional beams are updated in time to provide the desired power gain. To tackle this issue, this paper proposes a federated learning (FL) aided beam selection framework, where multiple ground base stations (GBSs) collaboratively train a global network. Simulation results reveal that the proposed multimodal fusion approach outperforms the baseline method, i.e., the pre-trained ResNet50 model, by improving beam selection accuracy by 3.7% to 14.8%. Under constrained training data conditions, the proposed method demonstrates robust generalization capabilities. Additionally, compared to methods that do not utilize FL, the proposed approach achieves an accuracy improvement of 5.3% to 14.8%. The above observations from the simulation results underscore the effectiveness of integrating multimodal fusion with FL in enhancing the accuracy of network beam selection. The proposed framework shows strong potential for advancing communication applications in low-altitude UAV scenarios.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Multimodal Data Fusion for Beam Selection in UAV Communications with Federated Learning


    Beteiligte:
    Yang, Xiuyuan (Autor:in) / Zhou, Quan (Autor:in) / Zhao, Jingjing (Autor:in) / Cai, Kaiquan (Autor:in) / Zhu, Yanbo (Autor:in)


    Erscheinungsdatum :

    08.04.2025


    Format / Umfang :

    562198 byte





    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch




    Collaborative Multimodal Vehicular Transformer Training Using Federated Learning

    Cao, Xingjian / Li, Zonghang / Sun, Gang et al. | IEEE | 2024


    FMAD: Fusion-based Multimodal Abnormal Detection Scheme for Vehicular Communications

    Nguyen, Van-Linh / Nguyen, Lan-Huong / Ting, Hao-En | IEEE | 2024



    Federated Learning-Based Resource Allocation for V2X Communications

    Bhardwaj, Sanjay / Kim, Da-Hye / Kim, Dong-Seong | IEEE | 2025