This paper proposes an adaptive communication resource allocation algorithm to address the model transfer challenge in federated learning (FL). Our approach dynamically adjusts transmission power and channel coding rates based on FL training states, utilizing a bit-position-aware transmission scheme rooted in unequal error protection (UEP) principles. The system's performance is analytically assessed through a derived upper bound on the FL convergence rate, considering model transmission accuracy and client gradient diversities in heterogeneous data distributions. Then a resource allocation problem, aimed at minimizing the proposed upper bound, is decomposed into two sub-problems and solved. The first sub-problem, minimizing a derived upper bound on model transmission error, is tackled using a proposed algorithm for adaptive bit power allocation. The second sub-problem, minimizing cumulative gradient diversities, is formulated as a Markov decision process (MDP) and solved using deep Q-learning. Numerical evaluations show that our method outperforms vanilla and existing UEP-coded algorithms.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Adaptive Communication Resource Allocation for Federated Learning with UEP Strategies


    Contributors:
    Lan, Muhang (author) / Xiao, Song (author) / Zhang, Wenyi (author)


    Publication date :

    2024-06-24


    Size :

    1428314 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Privacy-Preserving Resource Allocation for Asynchronous Federated Learning

    Chen, Xiaojing / Zhou, Zheer / Ni, Wei et al. | IEEE | 2024


    Federated Learning-Based Resource Allocation for V2X Communications

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



    Federated Reinforcement Learning for Resource Allocation in V2X Networks

    Xu, Kaidi / Zhou, Shenglong / Li, Geoffrey Ye | IEEE | 2024


    Joint Partial Offloading and Resource Allocation for Vehicular Federated Learning Tasks

    Ma, Guifu / Hu, Manjiang / Wang, Xiaowei et al. | IEEE | 2024