Massive Multi-Antenna Communications with Low-Resolution Data Converters

Abstract: Massive multi-user (MU) multiple-input multiple-output (MIMO) will be a core technology in future cellular communication systems. In massive MU-MIMO systems, the number of antennas at the base station (BS) is scaled up by several orders of magnitude compared to traditional multi-antenna systems with the goals of enabling large gains in capacity and energy efficiency. However, scaling up the number of active antenna elements at the BS will lead to significant increases in power consumption and system costs unless power-efficient and low-cost hardware components are used. In this thesis, we investigate the performance of massive MU-MIMO systems for the case when the BS is equipped with low-resolution data converters. First, we consider the massive MU-MIMO uplink for the case when the BS uses low-resolution analog-to-digital converters (ADCs) to convert the received signal into the digital domain. Our focus is on the case where neither the transmitter nor the receiver have any a priori channel state information (CSI), which implies that the channel realizations have to be learned through pilot transmission followed by BS-side channel estimation, based on coarsely quantized observations. We derive a low-complexity channel estimator and present lower bounds and closed-form approximations for the information-theoretic rates achievable with the proposed channel estimator together with conventional linear detection algorithms. Second, we consider the massive MU-MIMO downlink for the case when the BS uses low-resolution digital-to-analog converters (DACs) to generate the transmit signal. We derive lower bounds and closed-form approximations for the achievable rates with linear precoding under the assumption that the BS has access to perfect CSI. We also propose novel nonlinear precoding algorithms that are shown to significantly outperform linear precoding for the extreme case of 1-bit DACs. Specifically, for the case of symbol-rate 1-bit DACs and frequency-flat channels, we develop a multitude of nonlinear precoders that trade between performance and complexity. We then extend the most promising nonlinear precoders to the case of oversampling 1-bit DACs and orthogonal frequency-division multiplexing for operation over frequency-selective channels. Third, we extend our analysis to take into account other hardware imperfections such as nonlinear amplifiers and local oscillators with phase noise. The results in this thesis suggest that the resolution of the ADCs and DACs in massive MU-MIMO systems can be reduced significantly compared to what is used in today's state-of-the-art communication systems, without significantly reducing the overall system performance.

  CLICK HERE TO DOWNLOAD THE WHOLE DISSERTATION. (in PDF format)