Analysis and Optimization of 3D Antenna Radiation Pattern in Cellular Networks

This thesis investigates the impact of 3D antenna radiation patterns on millimeter-wave (mm-Wave) cellular networks and heterogeneous cellular networks (HTCNs), and optimizes BS antenna downtilts to improve network performance.
Mm-Wave cellular networks and two-tier HTCNs with uniform rectangular arrays (URAs) are modeled to investigate the impacts of downtilts and BS statistical measures on downlink transmission performances. Results indicate that appropriate measure adjustments can significantly enhance the cell-averaged spectral efficiency (SE) of mm-Wave networks, and spatially-averaged coverage probability, area spectral efficiency (ASE), and energy efficiency (EE) of HTCNs. Additionally, beam-selection is applied to enhance the EE in mm-Wave networks with various precoding schemes. Analysis of HTCN cell-edge average coverage probability under specific BS deployments reveals negligible impact of downtilts and transmit power.
Optimizing antenna downtilts is critical for effective network deployment. In mm-Wave networks, optimal BS antenna heights and downtilts for cell-averaged SE maximization are initially obtained through quick numerical searches, providing useful deployment guidelines. A tractable iterative algorithm is then proposed for 3D beamforming optimization, jointly optimizing precoding and downtilting to maximize the sum SE. Results validate the algorithm's feasibility, convergence, and potential for significant sum SE gains. In HTCNs, the optimal downtilts across tiers are derived via partial derivatives and the bisection method, showing that the downtilt-pair optimization can greatly enhance the spatially-averaged coverage probability, ASE, and EE.
To facilitate fast computation of integrals in analytical expressions, a Riemann sum approximation method is proposed by discretizing user positions to approximate their continuous distribution in mm-Wave networks. Meanwhile, a Gauss-Chebyshev quadrature method is applied to transform definite integrals into weighted sums in HTCNs. Results validate the accuracy and effectiveness of these methods.
Future work could explore more complex transmission scenarios, advanced interference management techniques, and dynamic user behaviors to further expand the potential of 3D antenna pattern design in cellular networks.