Multi-server federated learning (FL) has emerged as a promising paradigm to alleviate the communication bottlenecks of single-server FL by exploiting edge-level aggregation. In realistic dense deployments, the coverage regions of neighboring edge servers (ESs) often overlap, enabling some clients to communicate with multiple ESs. However, existing multi-server FL schemes fail to fully leverage such overlapping clients for efficient inter-server collaboration, leading to excessive reliance on cloud aggregation and slow convergence under heterogeneous data distributions. To address this challenge, we propose FedOC, a novel multi-server FL framework that explicitly exploits overlapping clients to accelerate training. In FedOC, overlapping clients can serve as relay overlapping clients (ROCs) to enable real-time edge-to-edge model relaying, or as normal overlapping clients (NOCs) that dynamically select edge models for local training based on delivery latency, thereby facilitating indirect data fusion across ESs. By integrating decentralized inter-ES aggregation with latency-aware client training, FedOC significantly reduces the need for frequent cloud aggregation while improving training efficiency. Extensive experiments on MNIST, Fashion-MNIST and CIFAR-10 demonstrate that FedOC achieves substantially faster convergence and higher final accuracy than other baselines. In particular, under strong data heterogeneity and cloud-free aggregation, FedOC improves the final test accuracy by approximately 3%–25% on these datasets compared with other baselines, while significantly reducing the latency required to reach a target accuracy.

Multi-server Federated Learning (FL) has emerged as a promising approach to alleviate the communication bottle-necks of traditional single-server FL. In practical deployments, the coverage areas of different edge servers (ESs) may overlap, allowing clients in overlapping regions to access models from multiple ESs. Leveraging this observation, we enable overlapping clients (OCs) to relay edge models between neighboring ESs and dynamically select suitable models for local training, facilitating multi-hop model propagation across ESs without relying on frequent cloud aggregation. This design significantly reduces communication latency while improving training efficiency. We derive a convergence upper bound for the above OCs-based FL framework, which explicitly quantifies the impact of inter-server propagation on convergence error. Guided by this theoretical result, we formulate an optimization problem that aims to maximize dissemination range of each ES model among all ESs by OCs within a limited latency. To solve this problem, we develop a conflict-graph-based local search algorithm optimizing the routing strategy and scheduling the transmission times of individual ESs to its neighboring ESs. By integrating decentralized inter-ES aggregation with latency-aware client training, our proposed algorithm significantly reduces the need for frequent cloud aggregation while improving training efficiency. Extensive experimental results show remarkable performance gains of our scheme compared to existing state-of-the-art methods.

To address the challenges posed by the heterogeneity inherent in federated learning (FL) and to attract high-quality clients, various incentive mechanisms have been employed. However, existing incentive mechanisms are typically utilized in conventional synchronous aggregation, resulting in significant straggler issues. In this study, we propose a novel asynchronous FL framework that integrates an incentive mechanism based on contract theory. Within the incentive mechanism, we strive to maximize the utility of the task publisher by adaptively adjusting clients' local model training epochs, taking into account time delay and test accuracy. In the asynchronous scheme, considering client quality, we devise aggregation weights and an access control algorithm to facilitate asynchronous aggregation. Through experiments conducted on the MNIST dataset, our framework achieved a test accuracy that is 3.12% and 5.84% higher than the accuracy achieved by FedAvg and FedProx without any attacks, respectively. Under attacks, the framework exhibits a 1.35% accuracy improvement over the ideal Local SGD. Furthermore, aiming for the same target accuracy, our framework demands notably less computation time than both FedAvg and FedProx.

Over-the-air Computation (AirComp) has been demonstrated as an effective transmission scheme to boost the efficiency of federated edge learning (FEEL). However, existing FEEL systems with AirComp scheme often employ traditional synchronous aggregation mechanisms for local model aggregation in each global round, which suffer from the stragglers issues. In this paper, we propose a semi-asynchronous aggregation FEEL mechanism with AirComp scheme (PAOTA) to improve the training efficiency of the FEEL system in the case of significant heterogeneity in data and devices. Taking the staleness and divergence of model updates from edge devices into consideration, we minimize the convergence upper bound of the FEEL global model by adjusting the uplink transmit power of edge devices at each aggregation period. The simulation results demonstrate that our proposed algorithm achieves convergence performance close to that of the ideal Local SGD. Furthermore, with the same target accuracy, the training time required for PAOTA is less than that of the ideal Local SGD and the synchronous FEEL algorithm via AirComp.

Federated learning (FL) can train a global model from clients' local data set, which can make full use of the computing resources of clients and performs more extensive and efficient machine learning on clients with protecting user information requirements. Many existing works have focused on optimizing FL accuracy within the resource constrained in each individual round, however there are few works comprehensively consider the optimization for latency, accuracy and energy consumption over all rounds in wireless federated learning. Inspired by this, in this paper, we investigate FL in wireless networks where client selection and bandwidth allocation are two crucial factors which significantly affect the latency, accuracy and energy consumption of clients. We formulate the optimization problem as a mixed-integer problem, which is to minimize the cost of time and accuracy within the long-term energy constrained over all rounds. To address this optimization problem, we propose a per-round energy drift plus cost (PEDPC) algorithm from an online perspective, and the performance of the PEDPC algorithm is verified in simulation results in terms of latency, accuracy and energy consumption in IID and NON-IID data distributions.