Massive MIMO and Multi-hop Mobile Communication Systems

Since the late 1990s, massive multiple-input multiple-output (MIMO) has been suggested to improve the achievable data rate in wireless communication systems. To overcome the high path losses in the high frequency bands, the use of massive MIMO will be a must rather than an option in future wireless communication systems. At the same time, due to the high cost and high energy consumption of the traditional fully digital beamforming architecture, a new beamforming architecture is required. Among the proposed solutions, the hybrid analog digital (HAD) beamforming architecture has received considerable attention. The promising massive MIMO gains heavily rely on the availability of accurate channel state information (CSI). This thesis considers a wideband massive MIMO orthogonal frequency division multiplexing (OFDM) system. We propose a channel estimation method called sequential alternating least squares approximation (SALSA) by exploiting a hidden tensor structure in the uplink measurement matrix considering the HAD architecture at the base station (BS). Then, the channel matrix is estimated sequentially using an alternating least squares (ALS) method. The proposed SALSA algorithm is applicable for any massive MIMO system regardless of the channel characteristics. Detailed simulation results show that SALSA offers flexible control of the complexity-accuracy trade-off compared to the classical least squares (LS) and minimum mean squared error (MMSE) methods. Then, we propose a HAD beamforming with a low-complexity design applicable to any hybrid architecture, where we decouple the designing of the baseband and analog beamforming matrices. We show that the analog beamforming matrix design can be written as a series of convex subproblems that are updated iteratively until convergence is obtained. Compared to the reference algorithms, our framework either achieves a comparable performance or significantly outperforms them depending on the HAD beam- forming architecture. Our proposed HAD beamforming design has a low-complexity and is applicable for any hybrid architecture. Exploring new less-congested frequency bands, such as millimeter wave (mmWave) frequencies, i.e., 30 – 300 GHz, has been proposed as a promising solution to increase network capacity. However, mmWave and high-frequency bands suffer from a high path loss compared to the sub-6 GHz band, which renders them applicable only for short-range and indoor communication scenarios. Reconfigurable intelligent surfaces (RISs) have been introduced as a low-cost and energy-efficient green solution to improve the communication range and overcome blockage issues in future wireless communication networks. This thesis considers an RIS-aided mmWave MIMO system, where the RIS is constituted of passive reflecting elements. We propose two novel channel estimation techniques for Single RIS (S-RIS)-aided systems. By exploiting the low-rank nature of mmWave channels in the angular domains, we propose a non-iterative Two-stage RIS-aided Channel Estimation (TRICE) framework, where every stage is formulated as a multi-dimensional direction of arrival (DoA) estimation problem. As a result, our TRICE framework is very general in the sense that any efficient multi-dimensional DoA estimation solution can be readily used in every stage to estimate the associated channel parameters. Then, we extend our TRICE framework by exploiting the tensor structure of the received signals, which admit a canonical polyadic decomposition (CPD which is also known as Parallel Factor (PARAFAC) analysis. This extension is called Tensor-based RIS Channel Estimation (TenRICE), in which the tensor factor matrices are estimated via an ALS method. Numerical simulations evaluate the resulting system performance and show that both methods require a lower training overhead and a lower computational complexity, compared to the benchmark solutions, while TenRICE has a superior performance approaching the Cram?r Rao lower Bound (CRB). After that, we propose a heuristic non-iterative two-step method to design the RIS reflection coefficients, where the RIS reflection vector is obtained in a closed-form using the Frobenius-Norm Maximization (FroMax) strategy. Our simulation results show that FroMax achieves a comparable performance to benchmark methods with a lower complexity. Since the performance of the S-RIS-aided systems depends on the distance of the RIS to the transmitter and the receiver, multi-RIS-aided systems have attracted more attention in the recent years. This thesis considers a flat-fading Double RIS (D-RIS)-aided MIMO system and proposes channel estimation techniques when (i) both RISs have the same training overhead and (ii) RISs have different training overheads. For case (i), we propose an ALS-based channel estimation method, called channel estimation for joint training (CEJOINT), by exploiting the Tucker2 tensor structure of the received signals. We show that if the reflective elements of the S-RIS system are carefully distributed between the two RISs in a D-RIS system, the training overhead in the D-RIS system can be reduced and the estimation accuracy can also be increased compared to the S-RIS system. For case (ii), we show that the received signal can be represented as a 4-way tensor satisfying a nested PARAFAC decomposition model. Exploiting such a structure, we propose a closed-form Khatri-Rao Factorization (KRF)-based and an iterative ALS-based channel estimation method, which are called KRF-based for separate training (KRF-SEPT) and ALS-based for separate training (ALS-SEPT), respectively. The numerical results show that both proposed methods have a comparable performance as long as the identifiability conditions of the KRF-SEPT are satisfied, while the ALS-SEPT method can achieve a satisfactory performance with less training overhead. Moreover, the performance of the proposed ALS-SEPT method can further be improved by using KRF-SEPT as an initialization.