Log in
Back to feed

Similar papers

Conference Open access Jun 2026

Drift-Aware Straggler Mitigation for Heterogeneous Split Federated Learning via Proactive Cut-Layer and Asynchronous Server Updates

Split Federated Learning (SFL) alleviates the computation burden on resource-constrained clients by partitioning model training between clients and the server. However, conventional synchronous SFL is highly vulnerable to straggler clients under heterogeneous compute, network, and data conditions, leading to prolonged training time and degraded system efficiency. To address this issue, we propose DAPS-SFL, a drift-aware and proactive straggler mitigation framework for heterogeneous SFL. The proposed method integrates three components: a lightweight straggler risk predictor based on normalized runtime statistics, a per-client adaptive cut-layer policy that dynamically adjusts the split point according to estimated computation and communication costs, and a drift-aware asynchronous server update mechanism that jointly accounts for update staleness and distribution shift. Experiments on CIFAR-10 with 50 clients under Dirichlet non-IID settings and heterogeneous compute/network environments demonstrate that DAPS-SFL reduces the wall-clock time required to reach 70% accuracy by approximately 22% compared with synchronous SplitFed, improves final test accuracy to 76.8 ± 0.4%, and significantly lowers tail latency while incurring only modest communication overhead. These results indicate that jointly optimizing cut-layer adaptation and asynchronous aggregation is an effective direction for improving the robustness and efficiency of SFL in heterogeneous environments.

Junqi Zhang, Bingxu Chen, Yihang Tang et al. · 1 citation
2026

Distribution Deviation-Aware Split Federated Learning in Resource-Limited Wireless Networks

The escalating complexity of deep neural networks introduces substantial challenges to deploying federated learning (FL) in resource-limited edge environments. To address these limitations, split federated learning (SFL) has emerged as a promising paradigm, alleviating client-side computational and communication burdens via strategic model splitting, and periodically aggregating client-side and server-side models consistent with the principles of FL. Nevertheless, existing SFL frameworks encounter significant performance degradation arising from data heterogeneity and imbalance, client heterogeneity, as well as constrained wireless resources. To overcome these issues, this paper introduces a novel data distribution deviation-aware split federated learning (DA-SFL) framework. DA-SFL dynamically adjusts aggregation weights according to the deviation of clients’ data distributions from a global distribution, effectively mitigating biases induced by data imbalance and heterogeneity. Furthermore, we theoretically establish the convergence bound of DA-SFL under a non-convex loss function setting, demonstrating that minimizing the data deviation in each training round enhances learning efficacy. Motivated by this, we formulate a mixed-integer nonlinear programming to optimize learning performance under long-term energy constraints. Leveraging the Lyapunov optimization framework, we decompose the problem into a series of tractable subproblems in each learning round, and propose efficient algorithms to find the client scheduling, adaptive cut layer selection, bandwidth allocation, and aggregation weighting policies. Extensive experimental evaluations conducted on Fashion-MNIST, CIFAR-10, and CINIC-10 datasets across diverse scenarios of data heterogeneity and imbalance demonstrate that DA-SFL significantly outperforms baselines regarding test accuracy, time and energy efficiency, while exhibiting notable robustness and scalability.

Chunfeng Xie, Zhixiong Chen, Wenqiang Yi et al. · 1 citation
Conference Jul 2026

FedEdge-Adapt: Adaptive Federated Learning for Heterogeneous Edge AI Systems

Federated learning (FL) on heterogeneous edge networks faces a fundamental tension: standard aggregation protocols assume client homogeneity, yet real-world edge deployments span device tiers with $7 \times$ compute and $\mathbf{1 0} \times$ bandwidth disparities. Slow clients become stragglers that stall synchronous rounds, while uniform gradient compression degrades accuracy on bandwidth-constrained devices. This paper presents FedEdge-Adapt, a novel adaptive federated learning framework that jointly addresses device heterogeneity, straggler mitigation, and communication efficiency without sacrificing model quality. FedEdge-Adapt introduces three tightly coupled mechanisms: (1) tier-aware gradient compression that applies device-class-specific sparsification ratios, (2) drift-corrected aggregation that reweights client updates based on staleness and data heterogeneity, and (3) predictive client selection that anticipates dropout-prone devices using a lightweight resource oracle. We evaluate FedEdge-Adapt on a 30-node heterogeneous edge network over 150 communication rounds using the CIFAR-10 dataset under non-IID distributions ($\alpha=0.5$ Dirichlet) and compare against FedAvg, FedProx, and SCAFFOLD baselines. FedEdge-Adapt achieves 85.44% global accuracy, a 6.85 percentage-point improvement over FedAvg, while simultaneously reducing round latency by $\mathbf{6 7. 4 \%}$, communication overhead by $\mathbf{3 4. 0 \%}$, and client dropout rate by $\mathbf{5 4. 0 \%}$. Convergence is reached in 18 rounds versus 31 for FedAvg. Extended experiments across 100+ rounds confirm long-term stability with no late-stage divergence.

Saher Elsayed, Mohamed Ali, Samer Abubaker et al. · 0 citations
Preprint Jul 2026

Robust Federated Learning Under Real-World Client Churn

Federated Learning (FL) enables training shared models on private, on-device data, but production deployments remain constrained to slow, multi-day refresh cycles due to the complexity of coordinating massive client populations. For applications such as feed ranking, ad targeting, and personalized recommendation, model freshness: the ability to rapidly adapt to new user-local data is critical for maximizing objectives like click-through rate. This lag leaves models stale and unresponsive to volatile data distributions driven by viral trends and shifting user intent. Bridging this gap requires addressing three challenges overlooked by existing FL systems: transient client availability, dynamic data heterogeneity, and delays between model predictions and observable outcomes. We present FeLiX, an FL orchestration framework that minimizes wall-clock time-to-target accuracy on live interaction streams. FeLiX introduces three primitives: (i) streaming-aware availability tiers that leverage lightweight telemetry to identify ready clients at scale; (ii) fresh-utility selection, a dual-tier mechanism that prioritizes statistically valuable updates from devices able to meet tight refresh deadlines; and (iii) informativeness-aware, delay-robust aggregation that incorporates late, high-value updates containing ground-truth outcomes without biasing the global model toward stale distributions. Unlike prior systems that rely on unrealistic oracular knowledge of client availability, FeLiX achieves near-oracular performance in real-world settings. Across CIFAR-10, Google Speech, and realistic low-availability traces, FeLiX reduces wall-clock time-to-target accuracy by up to 2.37X while reducing communication bandwidth by 1.30X compared to state-of-the-art synchronous and asynchronous FL baselines.

Dhruv Garg, Neha Lakhani, Debopam Sanyal et al. · 0 citations
Open access Jul 2026

Federated Learning over 5G/6G Networks: Dynamic Client Selection and Resource Allocation for Heterogeneous Edge Environments

Federated learning (FL) has emerged as a promising paradigm for privacy-preserving edge intelligence because it enables geographically distributed devices to collaboratively train a shared model without transferring raw data to a central cloud. This capability is particularly valuable for 5G and emerging 6G networks, where edge-native services are required to satisfy stringent latency, bandwidth, and privacy constraints while operating on highly heterogeneous devices and time-varying wireless channels. In practice, however, synchronous FL is often constrained by straggling clients with limited computation capability or unfavorable communication conditions, which increases round latency and reduces overall resource efficiency. To address this challenge, this study develops a rigorously structured framework for dynamic client selection and radio resource allocation in heterogeneous wireless edge environments. Each FL round is formulated as a latency-aware scheduling problem that jointly captures local computation time, uplink transmission time, minimum participation constraints, and resource block assignment. On this basis, we propose a Dynamic Client Selection and Resource Allocation (DCS-RA) method that integrates computation-aware, channel-aware, and fairness-aware scoring with greedy resource block allocation guided by marginal completion time reduction. The study further provides a clear methodological structure, workflow visualization, literature-grounded justification, dataset documentation, and uncertainty-aware result reporting. Under the reported simulation setting with 100 clients and 20 resource blocks, DCS-RA reduces the average round completion time from 1.92 s to 1.55 s on MNIST and from 2.02 s to 1.57 s on CIFAR-10, corresponding to improvements of 19.39% and 22.47%, respectively. Standard deviation reductions of 70.59% and 80.77% further indicate improved round-to-round stability and more reliable training behavior. These results support the central conclusion that lightweight joint scheduling can materially improve wall-clock FL efficiency in heterogeneous 5G/6G edge networks.

A. Al-Karawi, R. Akdeniz · 0 citations
2026

When Split Federated Learning Meets Prototype Learning: A Communication-Efficient Approach in Wireless Networks

Nowadays, split federated learning (SFL) has emerged as an effective paradigm for enabling privacy-preserving collaborative intelligence across heterogeneous devices with limited computation. However, SFL incurs significant communication overhead in wireless networks due to the uplink transmission of high-dimensional smashed data, which degrades network efficiency. To mitigate the communication bottleneck, we propose a prototype-based SFL framework ProtoSFL. Specifically, each selected client computes local prototypes for observed classes and uploads them to the server. Based on the received prototypes, the server derives global prototypes and optimizes a weighted objective that combines classification loss with prototype alignment loss. The server then updates the model accordingly and returns personalized prototype gradients to the clients. Simulation results verify the effectiveness of ProtoSFL in reducing communication overhead, achieving a substantial reduction in uplink communication, while maintaining competitive testing accuracy under various heterogeneous data settings compared with SFL baselines.

Xinran Zhang, Xianke Qiang, Weilong Chen et al. · 0 citations