Federated learning improves wireless training without complex signal syncing
Non-Coherent Over-the-Air Federated Learning: Protocol, Convergence, and Device Scheduling
Information TheoryMachine Learning
Summary
Getting many devices to work together on training AI models over wireless networks is tricky because their signals must line up perfectly, which is hard and slow. The researchers propose a method that avoids needing exact signal timing or channel details by using special techniques like binary dithering and error feedback. Their method still learns effectively and nearly as fast as the ideal but complicated approaches. They also suggest ways to pick and control devices to speed up learning when network conditions vary. Tests with image data sets show their method works well in real-world settings.
federated learningover-the-air aggregationmultiple-access channelchannel state informationnon-coherent detectionbinary ditheringdevice schedulingconvergence rateFedAvgwireless communication
Authors
Haifeng Wen, Nicolò Michelusi, Osvaldo Simeone, Yang Yang, Hong Xing
Abstract
To mitigate the scalability bottleneck in the radio access network (RAN) in federated edge learning (FEEL), over-the-air federated learning (AirFL) exploits waveform superposition over multiple-access channels (MACs) for analog model aggregation. However, coherent AirFL typically relies on stringent PHY-layer conditions such as accurate channel state information (CSI), tight time/frequency synchronization, and frequent transceiver calibration for signal alignment. However, these requirements, if not impossible to be met, incur substantial communication and computation overhead. In this paper, we propose a non-coherent AirFL (NCAirFL) protocol over a broadband single-antenna MAC, leveraging binary dithering, unbiased non-coherent detection, and long-term error feedback to waive the need for instantaneous CSI. For NCAirFL with general smooth non-convex objectives and a constant learning rate, we establish a convergence bound achieving the convergence rate in the same order of $\mathcal{O}(1/\sqrt{T})$ as communication-ideal FedAvg, where $T$ is the total number of communication rounds. To further improve communication efficiency under data and wireless resource heterogeneity, we also derive a lower bound on the expected single-round objective decrease in the global loss conditioned on device scheduling, building upon which a surrogate objective function is obtained for jointly optimal device selection and power control. Experimental results on MNIST and CIFAR-10 corroborate that NCAirFL achieves learning performance close to FedAvg in practical settings, with the proposed device scheduling policy substantially accelerating convergence.