Quantized MIMO: Channel Capacity and Spectrospatial Power Distribution

TL;DR

This paper analyzes the impact of low-resolution quantization on MIMO systems at millimeter wave frequencies, providing an information-theoretic model for capacity and power distribution that is validated through simulations.

Contribution

It introduces a linear Gaussian noise model to accurately describe quantization effects on MIMO capacity and power distribution, simplifying analysis and design.

Findings

The model accurately predicts spectral and spatial power distributions.

A lower bound on system capacity under quantization is derived.

Validation confirms the model's effectiveness in practical scenarios.

Abstract

Millimeter wave systems suffer from high power consumption and are constrained to use low resolution quantizers --digital to analog and analog to digital converters (DACs and ADCs). However, low resolution quantization leads to reduced data rate and increased out-of-band emission noise. In this paper, a multiple-input multiple-output (MIMO) system with linear transceivers using low resolution DACs and ADCs is considered. An information-theoretic analysis of the system to model the effect of quantization on spectrospatial power distribution and capacity of the system is provided. More precisely, it is shown that the impact of quantization can be accurately described via a linear model with additive independent Gaussian noise. This model in turn leads to simple and intuitive expressions for spectrospatial power distribution of the transmitter and a lower bound on the achievable rate of the system. Furthermore, the derived model is validated through simulations and numerical evaluations, where it is shown to accurately predict both spectral and spatial power distributions.