Asymptotic SEP Analysis and Optimization of Linear-Quantized Precoding in Massive MIMO Systems

TL;DR

This paper analyzes and optimizes linear-quantized precoding in massive MIMO systems with low-resolution DACs, providing asymptotic performance insights and a new optimal precoder design.

Contribution

It offers a rigorous asymptotic analysis of SEP and SINR for linear-quantized precoding and derives an optimal precoder within a broad class of such schemes.

Findings

Asymptotic analysis accurately predicts finite system performance.

The derived precoder outperforms existing schemes.

The analysis justifies heuristic Bussgang-based approaches.

Abstract

A promising approach to deal with the high hardware cost and energy consumption of massive MIMO transmitters is to use low-resolution digital-to-analog converters (DACs) at each antenna element. This leads to a transmission scheme where the transmitted signals are restricted to a finite set of voltage levels. This paper is concerned with the analysis and optimization of a low-cost quantized precoding strategy, referred to as linear-quantized precoding, for a downlink massive MIMO system under Rayleigh fading. In linear-quantized precoding, the signals are first processed by a linear precoding matrix and subsequently quantized component-wise by the DAC. In this paper, we analyze both the signal-to-interference-plus-noise ratio (SINR) and the symbol error probability (SEP) performances of such linear-quantized precoding schemes in an asymptotic framework where the number of transmit antennas and the number of users grow large with a fixed ratio. Our results provide a rigorous justification for the heuristic arguments based on the Bussgang decomposition that are commonly used in prior works. Based on the asymptotic analysis, we further derive the optimal precoder within a class of linear-quantized precoders that includes several popular precoders as special cases. Our numerical results demonstrate the excellent accuracy of the asymptotic analysis for finite systems and the optimality of the derived precoder.