HybridFLow: SDN-Orchestrated Client Partitioning for Hybrid Federated Learning
HybridFLow uses SDN topology visibility to partition federated-learning clients into sync/async groups, reaching 80% accuracy 33-40% faster than SmartFLow.
HybridFLow is a closed-loop, SDN-driven orchestration framework for hybrid federated learning that integrates network-layer intelligence into cross-silo training. It leverages the SDN controller's global topology view to generate calibrated per-client communication-time estimates, partitioning clients into synchronous and asynchronous groups while balancing round latency and update staleness, with measured times fed back after each round. Across multiple network topologies it reaches 80% target accuracy 33-40% faster than SmartFLow and cuts average round duration by 30-40 seconds, while FedAsync fails to reach target accuracy under non-IID data.
- Uses the SDN controller's global topology view for per-client communication-time estimates before each round.
- Measured communication times feed back to the controller to refine future predictions in a closed loop.
- Targets the straggler effect in wide-area cross-silo federated learning without sharing raw data.
- Reaches 80% accuracy 33-40% faster than SmartFLow; FedAsync fails under non-IID distributions.
Full article173 words · extracted from arxiv.org · click to collapse
Cross-silo Federated Learning (FL) enables geographically distributed institutions to collaboratively train machine learning models without sharing raw data. In wide-area deployments, however, communication delays often dominate round completion time and exacerbate the straggler effect. Hybrid FL addresses this challenge by combining synchronous and asynchronous client participation, but effective partitioning requires visibility into network conditions such as shared bottlenecks, link utilization, and path contention that individual clients cannot observe. We present HybridFLow, a closed-loop SDN-driven orchestration framework that integrates network-layer intelligence directly into hybrid FL. Leveraging the SDN controller's global topology view, HybridFLow generates calibrated per-client communication-time estimates before each training round and uses them to partition clients into synchronous and asynchronous groups while balancing round latency and update staleness. After each round, measured communication times are fed back to the controller to continuously refine future predictions. Experimental results across multiple network topologies show that HybridFLow reaches 80% target accuracy 33-40% faster than SmartFLow and reduces average round duration by 30-40 seconds, while FedAsync fails to reach the target accuracy under non-IID data distributions.
Text extracted automatically; images, tables and formatting may be missing. Original: https://arxiv.org/abs/2609.10404