Abstract:Federated Learning (FL) provides a privacy-preserving framework for training machine learning models on mobile edge devices. Traditional FL algorithms, e.g., FedAvg, impose a heavy communication workload on these devices. To mitigate this issue, Hierarchical Federated Edge Learning (HFEL) has been proposed, leveraging edge servers as intermediaries for model aggregation. Despite its effectiveness, HFEL encounters challenges such as a slow convergence rate and high resource consumption, particularly in the presence of system and data heterogeneity. However, existing works are mainly focused on improving training efficiency for traditional FL, leaving the efficiency of HFEL largely unexplored. In this paper, we consider a two-tier HFEL system, where edge devices are connected to edge servers and edge servers are interconnected through peer-to-peer (P2P) edge backhauls. Our goal is to enhance the training efficiency of the HFEL system through strategic resource allocation and topology design. Specifically, we formulate an optimization problem to minimize the total training latency by allocating the computation and communication resources, as well as adjusting the P2P connections. To ensure convergence under dynamic topologies, we analyze the convergence error bound and introduce a model consensus constraint into the optimization problem. The proposed problem is then decomposed into several subproblems, enabling us to alternatively solve it online. Our method facilitates the efficient implementation of large-scale FL at edge networks under data and system heterogeneity. Comprehensive experiment evaluation on benchmark datasets validates the effectiveness of the proposed method, demonstrating significant reductions in training latency while maintaining the model accuracy compared to various baselines.

Enhancing Decentralized and Personalized Federated Learning with Topology Construction

Topology Learning for Heterogeneous Decentralized Federated Learning over Unreliable D2D Networks

Decentralized Directed Collaboration for Personalized Federated Learning

Decentralized Federated Learning with Data Feature Transmission and Neighbor Selection.

Secure and Efficient Decentralized Federated Learning with Data Representation Protection

Towards More Suitable Personalization in Federated Learning via Decentralized Partial Model Training

Tensor Decomposition Based Personalized Federated Learning

Decentralized Federated Learning with Intermediate Results in Mobile Edge Computing

Topology Construction with Minimum Total Time for Geo-Distributed Decentralized Federated Learning.

Accelerating Decentralized Federated Learning in Heterogeneous Edge Computing

Like Attracts Like: Personalized Federated Learning in Decentralized Edge Computing

Towards Practical Overlay Networks for Decentralized Federated Learning

Enhancing Decentralized Federated Learning with Model Pruning and Adaptive Communication

Heterogeneity-Aware Resource Allocation and Topology Design for Hierarchical Federated Edge Learning

Adaptive Configuration for Heterogeneous Participants in Decentralized Federated Learning

Topology-Driven Synchronization Interval Optimization for Latency-Constrained Geo-Decentralized Federated Learning

Optimizing Communication Topology for Collaborative Learning Across Datacenters

Beyond the Federation: Topology-aware Federated Learning for Generalization to Unseen Clients

Accelerating Federated Edge Learning Via Topology Optimization

Enhancing Federated Learning Convergence with Dynamic Data Queue and Data Entropy-driven Participant Selection