Random Orthogonalization for Federated Learning in Massive MIMO Systems

TL;DR

This paper introduces random orthogonalization, a novel uplink communication method for federated learning in massive MIMO systems, leveraging channel hardening and favorable propagation to enable efficient over-the-air model aggregation without transmitter channel state information.

Contribution

The paper presents a new random orthogonalization technique that couples federated learning with massive MIMO properties, reducing channel estimation overhead and enabling natural model aggregation.

Findings

Achieves over-the-air model aggregation without transmitter channel state info.

Reduces channel estimation overhead at the receiver.

Establishes a relationship between convergence rate, number of clients, and antennas.

Abstract

We propose a novel uplink communication method, coined random orthogonalization, for federated learning (FL) in a massive multiple-input and multiple-output (MIMO) wireless system. The key novelty of random orthogonalization comes from the tight coupling of FL model aggregation and two unique characteristics of massive MIMO - channel hardening and favorable propagation. As a result, random orthogonalization can achieve natural over-the-air model aggregation without requiring transmitter side channel state information, while significantly reducing the channel estimation overhead at the receiver. Theoretical analyses with respect to both communication and machine learning performances are carried out. In particular, an explicit relationship among the convergence rate, the number of clients and the number of antennas is established. Experimental results validate the effectiveness and efficiency of random orthogonalization for FL in massive MIMO.