Beamforming strategies in wireless multi-antenna communication systems

In this thesis, we study beamforming strategies for downlink transmission in different wireless communication systems. The major objective is to properly design the beamformer at the transmitter to mitigate the interference at the receivers such that the overall system performance can be enhanced, e.g., the network throughput is maximized. Firstly, the sum-rate optimization problem in a multicell system is considered. The original problem is transformed to a convex semidefinite programming problem by applying the uplink-downlink duality followed by semidefinite relaxation, which can be solved by using convex optimization techniques. Moreover, beamforming strategies in the massive multiple-input multiple-output (MIMO) system of the fifth-generation (5G) network are investigated. Assuming that the base station (BS) is equipped with two-dimensional (2D) massive antenna arrays, we analyze the condition in which the pilot contamination can be mitigated by applying the Von Mises Fisher (VMF) channel model and exploiting the property of the channel covariance matrix. It is shown that the pilot contamination can be mitigated if the angle of arrivals (AoAs) of the interfering users have a nonoverlapping region with the AoA of the desired user. We propose two realistic three-dimensional (3D) beamforming strategies, namely, region-partition beamforming when the BS has perfect channel state information (CSI) and statistic-based beamforming when the BS only knows the statistical knowledge of a two-tier clustered user distribution. The probability density function (pdf) of the two-tier user distribution is derived accordingly. When the full CSI is not available at the BS, a novel limited feedback scheme based on the compressive sensing (CS) technique is proposed to offer accurate channel knowledge. The channel is compressed through linear measurement at the receiver and fed back to the BS. The CS measurement is modelled as a quantization procedure in which quantization noises are generated according to the number of feedback bits. The original channel can be recovered with high accuracy by solving an optimization problem at the BS. Lastly, the secrecy sum-rate maximization problem in Gaussian wiretap channels is considered. To guarantee the secure communication in an interference wiretap channel, we apply a novel decomposition framework to transform the original secrecy sum-rate function to decoupled convex approximation. The decomposition framework is based on Taylor-series expansion and is able to preserve the convexity of the secrecy sum-rate function and linearize the nonconvex part. Following that, we propose a distributed iterative optimization algorithm where the BSs iteratively solve a sequence of convex optimization subproblems in parallel until the optimal beamformers are obtained. Numerical results are presented to validate the theoretical analysis and effectiveness of the proposed approaches.DOCTOR OF PHILOSOPHY (EEE