NCAirFL: CSI-Free Over-the-Air Federated Learning Based on Non-Coherent Detection

TL;DR

NCAirFL introduces a CSI-free over-the-air federated learning scheme that uses non-coherent detection, binary dithering, and error compensation to enable efficient, coherent-free model training over wireless channels.

Contribution

It proposes a novel CSI-free AirFL scheme based on non-coherent detection, achieving convergence without channel estimation or feedback, and demonstrates competitive performance.

Findings

Achieves convergence rate of O(1/√T) for non-convex objectives.

Demonstrates competitive performance compared to ideal and coherent benchmarks.

Uses binary dithering and long-term memory for error compensation.

Abstract

Over-the-air federated learning (FL), i.e., AirFL, leverages computing primitively over multiple access channels. A long-standing challenge in AirFL is to achieve coherent signal alignment without relying on expensive channel estimation and feedback. This paper proposes NCAirFL, a CSI-free AirFL scheme based on unbiased non-coherent detection at the edge server. By exploiting binary dithering and a long-term memory based error-compensation mechanism, NCAirFL achieves a convergence rate of order $\mathcal{O}(1/\sqrt{T})$ in terms of the average square norm of the gradient for general non-convex and smooth objectives, where $T$ is the number of communication rounds. Experiments demonstrate the competitive performance of NCAirFL compared to vanilla FL with ideal communications and to coherent transmission-based benchmarks.