Abstract
In massive multiple-input multiple-output (mMIMO) systems, the technique of using low-resolution analog-to-digital converters (ADCs) to quantize received signals can
significantly reduce the receiver hardware cost as well as circuit power consumption.
However, it incurs quantization noise in the digital domain which degrades the quality
of signal reception. In particular, for 1- to 3-bit resolution quantizers, communication
systems become very nonlinear and non-Gaussian. Quantization noise can be correlated to the received signal as well as the thermal noise. These are critical issues that
can fundamentally change the way of receiver design and optimization, and thus they
must be rigorously studied.
The first contribution of this thesis lies in the use of Hermite polynomials to study
the linear approximation model of low-resolution quantized mMIMO signal reception.
Unlike other linear approximation models in the literature that assume the quantization
noise to be white Gaussian and uncorrelated to the received signal, the proposed model,
termed second-order Hermite expansion (SOHE), uses the second-order Hermite kernel
to describe the quantization noise and the first-order Hermite kernel for the signal
linear response. Novel mathematical theorems are established to explain key stochastic
characteristics of the quantization noise. The SOHE theory results in an enhanced
LMMSE channel equalizer which considerably improves the equalization performance
particularly for mMIMO receivers using 2-bit quantizers.
Another key contribution lies in an intensive study of the well-known dither effect
of low-resolution quantizers, which explains the phenomenon of constructive noise in
the mMIMO signal reconstruction. This study leads to a fundamental rethinking of
mMIMO signal transmission strategy, operating signal-to-noise ratio (SNR) setup and
more fundamentally the threshold optimization for low-resolution signal quantization.
All of these provide theoretical support for the optimization of low-resolution quantized
mMIMO.
Finally, an outlook of major technical challenges and future direction of the lowresolution mMIMO research is presented.