Linear One-Bit Precoding in Massive MIMO: Asymptotic SEP Analysis and Optimization

TL;DR

This paper analyzes the asymptotic symbol error probability of linear one-bit precoding schemes in massive MIMO systems, providing accurate SEP predictions and deriving an optimal precoder within the class.

Contribution

It introduces an asymptotic analysis framework for linear one-bit precoding, including well-known schemes, and derives an optimal precoding scheme based on this analysis.

Findings

SEP converges to that of a scalar Gaussian noise model

Analytical SEP predictions are highly accurate

The derived precoder outperforms existing schemes

Abstract

This paper focuses on the analysis and optimization of a class of linear one-bit precoding schemes for a downlink massive MIMO system under Rayleigh fading channels. The considered class of linear one-bit precoding is fairly general, including the well-known matched filter (MF) and zero-forcing (ZF) precoding schemes as special cases. Our analysis is based on an asymptotic framework where the numbers of transmit antennas and users in the system grow to infinity with a fixed ratio. We show that, under the asymptotic assumption, the symbol error probability (SEP) of the considered linear one-bit precoding schemes converges to that of a scalar ``signal plus independent Gaussian noise'' model. This result enables us to provide accurate predictions for the SEP of linear one-bit precoding. Additionally, we also derive the optimal linear one-bit precoding scheme within the considered class based on our analytical results. Simulation results demonstrate the excellent accuracy of the SEP prediction and the optimality of the derived precoder.