Distributed Precoding Design for Multi-Group Multicasting in Cell-Free Massive MIMO

TL;DR

This paper introduces distributed precoding algorithms for multi-group multicasting in cell-free massive MIMO, optimizing beamforming with reduced signaling and training resources, and demonstrating significant performance improvements.

Contribution

It proposes novel distributed precoding schemes using uplink/downlink training that eliminate backhaul CSI exchange and improve multicasting performance.

Findings

Distributed methods outperform local CSI-based designs.

Elimination of backhaul signaling for CSI exchange.

Effective in resource-scarce training scenarios.

Abstract

We consider multi-group multicast precoding designs for cell-free massive multiple-input multiple-output (MIMO) systems. To optimize the transmit and receive beamforming strategies, we focus on minimizing the sum of the maximum mean squared errors (MSEs) over the multicast groups, which is then approximated with the sum MSE to simplify the computation and signaling. We adopt an iterative bi-directional training scheme with uplink and downlink precoded pilots to cooperatively design the multi-group multicast precoders at each base station and the combiners at each user equipment in a distributed fashion. An additional group-specific uplink training resource is introduced, which entirely eliminates the need for backhaul signaling for channel state information (CSI) exchange. We also propose a simpler distributed precoding design based solely on group-specific pilots, which can be useful in the case of scarce training resources. Numerical results show that the proposed distributed methods greatly outperform conventional cell-free massive MIMO precoding designs that rely solely on local CSI.