Zero-Forcing Per-Group Precoding (ZF-PGP) for Robust Optimized Downlink Massive MIMO Performance

TL;DR

This paper introduces ZF-PGP, a new precoding method for massive MIMO systems that offers high performance and simplified implementation, especially effective with hybrid precoding and in correlated channels, while maintaining robustness to errors.

Contribution

The paper proposes a combined ZF-PGP precoding technique that balances high gains with low complexity, suitable for massive MIMO and hybrid precoding scenarios, improving over existing methods.

Findings

ZF-PGP outperforms ZF and RZF in simulations.

The method maintains high performance with channel estimation errors.

It enables independent decoding at UEs without interference.

Abstract

In this paper, we propose a new, combined Zero-Forcing Per-Group Precoding (ZF-PGP) method that achieves very high gains in comparison to ZF Precoding techniques, while it simultaneously offers individually separate streams to reach individual User Equipment (UE), i.e., it obliterates the need for coordinated, joint decoding by the group's UEs. Although ZF-PGP in general experiences a performance loss in comparison to the near-optimal Per-Group Precoding within groups (PGP-WG), it is much simpler to implement than PGP-WG. The method works extremely well in conjunction with Combined Frequency and Space Division and Multiplexing (CFSDM) in correlated channels or in cases in which the group's total number of antennas is greater than the group's available Virtual Channel Model Beams (VCMB) number. As we show by multiple examples, for a Uniform Linear Array (ULA), by combining the proposed technique with massive Multiple-Input Multiple-Output (MIMO) and Joint Spatial Division and Multiplexing for Finite Alphabets (JSDM-FA) type hybrid precoding or CFSDM a very efficient, near-optimal design with reduced complexity at the UEs is achieved, while each UE decodes its received data independently, without interference from data destined for other UEs. Due to JSDM-FA Statistical Channel State Information (SCSI) inner beamforming, the groups are easily identified. We show that the combined ZF-PGP scheme achieves very high gains in comparison to ZF and Regularized ZF (RZF) precoding and other modified sub-optimal techniques investigated, while its complexity is still relatively low due to PGP-WG. We also demonstrate the robustness of the proposed precoding techniques to channel estimation errors, showing minimal performance loss.