FedSAP: Federated Learning with Structured Adaptive Partitioning for Multi-Domain Heterogeneous Edge Devices

2026-10-01

AI Summary

This study addresses the challenges of resource heterogeneity and data domain shift in edge federated learning, where uniform compression discards high-value channels and aggregation mixes sensitive updates. To this end, we propose FedSAP, a novel framework that introduces the first domain-aware heterogeneous architecture, reformulating structured pruning as a budget-constrained ternary channel allocation (global/private/discard). By inferring pseudo-domains via gradient similarity, FedSAP integrates domain-guided allocation with type-matched aggregation to enable stable feature sharing while isolating domain-sensitive updates. Extensive experiments on the Digits and Office-Caltech datasets demonstrate that FedSAP achieves average accuracies of 76.00% and 72.67%, respectively, outperforming the strongest baselines while supporting client pruning rates of up to 80%.

Abstract

Federated learning (FL) on heterogeneous edge devices must jointly accommodate unequal resource budgets and domain-shifted local data. Existing resource-adaptive methods decide how much of a model each client trains but not where retained capacity should reside or how it should be shared, whereas federated domain-generalization methods usually assume a shared full architecture. Uniform compression can therefore discard high-utility channels, and a single aggregation path can mix transferable features with domain-sensitive updates. We propose FedSAP, a domain-aware heterogeneous FL framework that casts structured pruning as budget-constrained tri-state channel allocation. FedSAP converts each keep ratio into non-uniform layer budgets, assigns stable channels to a Global pool, useful domain-sensitive channels to pseudo-domain-specific Private pools, and low-utility channels to a Dropped state. This partition lets broadly useful features benefit from cross-client pooling while isolating domain-sensitive updates from incompatible clients. Domain-Guided Assignment infers pseudo-domains from shallow-gradient similarity, while Type-Matched Aggregation restricts each channel to its intended sharing scope. Across three random seeds, FedSAP reaches 76.00% and 72.67% mean global accuracy on Digits and Office-Caltech, exceeding the strongest baseline by 1.70 and 4.92 percentage points while supporting client pruning ratios of up to 80% across heterogeneous clients.

Problem

• Federated Learning
• Heterogeneous Edge Devices
• Domain Shift
• Structured Pruning
• Resource Adaptation

Innovation

• Federated Learning
• Structured Pruning
• Domain Generalization
• Heterogeneous Edge Devices
• Channel Allocation