A key potential of massive multiple-input multiple-output (MIMO) systems which has made it interesting from a practical standpoint is its ability to substantially increase the network capacity with inexpensive, low-power components. Nevertheless, the static power consumption at the base station (BS) will increase proportionally to the number of antennas. Hence, considering hardware-aware design together with power consumption at the BS seems necessary in realizing practical massive MIMO systems. Among the various components responsible for power dissipation at the BS, the contribution of analog-to-digital converters (ADCs) is known to be dominant. Consequently, the idea of replacing the high-power high-resolution ADCs with power efficient low-resolution ADCs could be a viable approach to address power consumption concerns at the massive MIMO BSs. In this thesis, we study two different architectures in design of massive MIMO systems with one-bit ADCs, namely, mixed-ADC architecture and spatial Σ∆ architecture. The basic premise behind the mixed-ADC architecture is to achieve the benefits of conventional massive MIMO systems by just exploiting small pairs of high-resolution ADCs. In spatial Σ∆ architecture, by subtracting the quantized output of one antennas radio frequency (RF) chain from the signal at an adjacent antenna, coupled with spatial oversampling, the quantization noise is shaped to angular regions that the signal of interest is not present. We start with finding the optimal distribution of high-resolution and one-bit ADCs in a base station with energy constraint to maximize the spectral efficiency. Then, we use the Bussgang decomposition to develop a linear minimum mean-squared error (LMMSE) channel estimator for mixed-ADC architecture based on the combined round-robin measurements and we derive a closed-form expression for the resulting mean-squared error (MSE). We also perform a spectral efficiency (SE) analysis of the mixed-ADC implementation for the maximum ratio combining (MRC) and zero-forcing (ZF) receivers, and obtain expressions for a lower bound on the SE that takes into account the channel estimation error and the loss of efficiency due to the round-robin training. Finally, the possible SE improvement that can be achieved by using an antenna selection algorithm is investigated. Next we introduce the spatial Σ∆ architecture by adopting the oversampling and Σ∆ quantization approach in the time domain signal processing. We propose the appropriate design of spatial Σ∆ architecture by applying a scalar version of Bussgang approach. The results of the analysis indicate the significant gain of the Σ∆ approach compared with standard one-bit quantization for users that lie in the angular sector where the shaped quantization error spectrum is low. To alleviate the impact of strong interference in systems with one-bit quantizer, we extend the Σ∆ approach to present spatial feedback beamformer (FBB) Σ∆ architecture. We compare the symbol error rate of the FBB Σ∆ array with that of a system with high-resolution ADCs. The results show the superior performance of the one-bit FBB Σ∆ architecture which achieves performance equivalent to that of a system with only high resolution ADCs. To complete our analysis, we study the impact of mutual coupling and space-constrained arrays on the performance of mixed-ADC and spatial Σ∆ architectures. This phenomenon can eliminate the need for round-robin training for channel estimation in mixed-ADC architectures. For spatial Σ∆ architecture which relies on spatial oversampling (antenna spacing less than half a wavelength), the impact of mutual coupling may become significant as the antenna spacing decreases. Unlike temporal oversampling, there is a limit to the amount of spatial oversampling that can be achieved, due to the physical dimensions of the antennas. In the last section of this dissertation, we show that the one-bit Σ∆ array is particularly advantageous in space-constrained scenarios, and can still provide significant gains even in the presence of mutual coupling when the antennas are closely spaced. Through this thesis, we show that by exploiting the advanced capabilities of (MIMO) signal processing methods, performance of massive MIMO systems with coarse quantization can be significantly improved.

Multiple-user (MU) multiple-input multiple-output (MIMO) technology, which involves using multiple antennas to simultaneously serve multiple users or devices, is a cornerstone of both 5G and 6G. The use of MIMO in ultra-dense networks with smaller cell sizes and more antennas will result in a proportional increase in both inter- and intra-cell interference. To manage the interference, precoding or beamforming is needed to steer the transmit signals towards intended users and mitigate interference.

Symbol level precoding (SLP) techniques exploit information about the symbols to be transmitted in addition to the channel state information (CSI), which can significantly improve performance at the expense of increased complexity at the transmitter. The additional degrees of freedom (DoF) provided by the symbol-level information make it possible to exploit the constructive component of the interference, converting it into constructive interference (CI) that can move the received signals further from the decision thresholds of the constellation points. CI-based SLP recasts the traditional viewpoint of interference as a source of degradation to one where interference is a potential resource that can be exploited.

In this dissertation, we firstly study the use of SLP in the downlink of a multiuser multiple-input-single-output (MU-MISO) cognitive radio (CR) network, where a primary base station (PBS) serving primary users (PUs) and a cognitive base station (CBS) serving cognitive users (CUs) share the same frequency band. The SLP approach is designed using the symbol-wise Maximum Safety Margin (MSM) criterion, which exploits the constructive multiuser interference present in such a network. We adapt the non-linear MSM precoder to both underlay and overlay CR scenarios, depending on whether or not the primary system shares its information with the cognitive system. Secondly, we investigate robust SLP designs in an overlay CR network, where the primary and secondary networks transmit signals concurrently, however, the PBS shares imperfect CSI with the CBS. We propose robust SLP schemes in this scenario and consider two different CSI error models. For the norm-bounded CSI error model, we adopt a max-min philosophy to conservatively achieve robust SLP constraints; for the additive quantization noise model (AQNM), we employ a stochastic constraint to formulate the problem. Simulation results show that, rather than simply trying to eliminate the network's cross-interference, the proposed robust SLP schemes enable the primary and secondary networks to aid each other in meeting their quality of service constraints. Moreover, we propose precoding design in multi-antenna systems with improper Gaussian interference (IGI), characterized by correlated real and imaginary parts. We first study block level precoding (BLP) and SLP assuming the receivers apply a pre-whitening filter to decorrelate and normalize the IGI. We then shift to the scenario where the base station (BS) incorporates the IGI statistics in the SLP design, which allows the receivers to employ a standard detection algorithm without pre-whitenting. Finally we address the case where the non-circularity of the IGI is unknown, and we formulate robust BLP and SLP designs that minimize the worst case performance in such settings. Interestingly, we show that for BLP, the worst-case IGI is in fact proper, while for SLP the worst case occurs when the interference signal is maximally improper, with fully correlated real and imaginary parts. The numerical results reveal the superior performance of SLP in terms of symbol error rate (SER) and energy efficiency (EE), especially for the case where there is uncertainty in the non-circularity of the jammer.