On the performance of hybrid beamforming for millimeter wave wireless networks

The phenomenal growth in the demand for mobile wireless data services is pushing the boundaries of modern communication networks. Developing new technologies that can provide unprecedented data rates to support the pervasive and exponentially increasing demand is therefore of prime importance in wireless communications. In existing communication systems, physical layer techniques are commonly used to improve capacity. Nevertheless, the limited available resources in the spectrum are unable to scale up, fundamentally restricting further capacity increase. Consequently, alternative approaches which exploit both unused and underutilised spectrum bands are highly attractive. This thesis investigates the use of the millimeter wave (mmWave) spectrum as it has the potential to provide unlimited bandwidth to wireless communication systems. As a first step toward realising mmWave wireless communications, a cloud radio access network using mmWave technology in the fronthaul and access links is proposed to establish a feasible architecture for deploying mmWave systems with hybrid beamforming. Within the context of a multi-user communication system, an analytical framework of the downlink transmission is presented, providing insights on how to navigate across the challenges associated with high-frequency transmissions. The performance of each user is measured by deriving outage probability, average latency and throughput in both noise-limited and interference-limited scenarios. Further analysis of the system is carried out for two possible user association configurations. By relying on large antenna array deployment in highly dense networks, this architecture is able to achieve reduced outages with very low latencies, making it ideal to support a growing number of users. The second part of this work describes a novel two-stage optimisation algorithm for obtaining hybrid precoders and combiners that maximise the energy efficiency (EE) of a general multi-user mmWave multiple-input, multiple-output (MIMO) interference channel network involving internet of things (IoT) devices. The hybrid transceiver design problem considers both perfect and imperfect channel state information (CSI). In the first stage, the original non-convex multivariate EE maximization problem is transformed into an equivalent univariate problem and the optimal single beamformers are then obtained by exploiting the correlation between parametric and fractional programming problems and the relationship between weighted sum rate (WSR) and weighted minimum mean squared error (WMMSE) problems. The second stage involves the use of an orthogonal matching pursuit (OMP)-based algorithm to obtain the energy-efficient hybrid beamformers. This approach produces results comparable to the optimal beam-forming strategy but with much lower complexity, and further validates the use of mmWave networks in practice to support the demand from ubiquitous power-constrained smart devices. In the third part, the focus is on the more practical scenario of imperfect CSI for multi-user mmWave systems. Following the success of hybrid beamforming for mmWave wireless communication, a non-traditional transmission strategy called Rate Splitting (RS) is investigated in conjunction with hybrid beamforming to tackle the residual multi-user interference (MUI) caused by errors in the estimated channel. Using this technique, the transmitted signal is split into a common message and a private message with the transmitted power dynamically divided between the two parts to ensure that there is interference-free transmission of the common message. An alternating maximisation algorithm is proposed to obtain the optimal common precoder. Simulation results show that the RS transmission scheme is beneficial to multi-user mmWave transmissions as it enables remarkable rate gains over the traditional linear transmission methods. Finally, the fourth part analyses the spectral efficiency (SE) performance of a mmWave system with hybrid beamforming whilst accounting for real-life practice transceiver hardware impairments. An investigation is conducted into three major hardware impairments, namely, the multiplicative phase noise (PN), the amplified thermal noise (ATN) and the residual additive transceiver hardware impairments (RATHI). The hybrid precoder is designed to maximise the SE by the minimisation of the Euclidean distance between the optimal digital precoder and the noisy product of the hybrid precoders while the hybrid combiners are designed by the minimisation of the mean square error (MSE) between the transmitted and received signals. Multiplicative PN was found to be the most critical of the three impairments considered. It was observed that the additive impairments could be neglected for low signal-to-noise-ratio (SNR) while the ATNs caused a steady degradation to the SE performance