Thesis Examination Committee
Prof Lancelot Fitzgerald JAMES, ISOM/HKUST (Chairperson)
Prof Khaled BEN LETAIEF, ECE/HKUST (Thesis Supervisor)
Prof Ali GHRAYEB, Department of Electrical and Computer Engineering, Texas A&M University at Qatar (External Examiner)
Prof Danny TSANG, ECE/HKUST
Prof Chin-Tau LEA, ECE/HKUST
Prof Qian ZHANG, CSE/HKUST
Uplifting the carrier frequency to the millimeter wave (mm-wave) band stands out as an effective approach to meet the capacity demands in 5G networks, as it provides orders of magnitude greater spectrum than current cellular bands. Large-scale antenna arrays are needed to provide effective beamforming gains to compensate the signal attenuation for mm-wave communications, which, however, brings formidable challenges to performance analysis, algorithm design, and hardware implementation. Conventional fully digital precoding techniques are inapplicable, as they require a separate radio frequency (RF) chain for each antenna element. Hybrid precoding was recently proposed as a cost-effective alternative, which requires a small number of RF chains and thus can significantly reduce the hardware cost and power consumption.
First, by leveraging a single RF chain, analog beamforming serves as an initial solution. While the optimal analog beamforming strategy can be readily determined, it is difficult to characterize the performance of mm-wave networks with analog beamforming, due to the complex intercell interference. Using tools from stochastic geometry, we propose an analytical framework for performance analysis of mm-wave networks with arbitrary antenna patterns, based on which a comprehensive investigation on the impact of directional antenna arrays in mm-wave networks is carried out. It is shown that the coverage probabilities of mm-wave networks increase as a non-decreasing concave function with the antenna array size.
To further improve the spectral efficiency, we then consider the hybrid precoding which supports spatial multiplexing. We present several proposals of hybrid precoding structures, focusing on three key aspects: spectral efficiency, computational efficiency, and hardware efficiency. Furthermore, a series of effective algorithms specialized for different hybrid precoder structures are proposed based on sophisticated techniques, e.g., manifold optimization, semidefinite relaxation, and binary optimization. Simulation results demonstrate the effectiveness of both the proposed hybrid precoder structures and hybrid precoding algorithms. Promising candidates for different application scenarios are identified through comprehensive comparisons between different hybrid precoding proposals, and key design insights are unraveled.