Interference management in 5G cellular networks

Publication Type:
Thesis
Issue Date:
2016
Full metadata record
Files in This Item:
Filename Description Size
01front.pdf140.07 kB
Adobe PDF
02whole.pdf1.21 MB
Adobe PDF
This dissertation is concerned with the nonconvex optimization problems of interference management under the consideration of new disruptive technologies in the fifth-generation cellular networks. These problems are the key to the successful roll-out of these new technologies but have remained unsolved due to their mathematical challenge. Therefore, this dissertation provides novel minorants/majorants of the nonconvex functions which are then used for the successive convex approximation framework. The first considered technology is heterogeneous networks (HetNet) in which base stations (BSs) of various sizes and types are densely deployed in the same area. Although HetNet provides a significant improvement in spectral efficiency and offloading, designing an optimal power transmission and association control policy is challenging, especially when both quality-of-service (QoS) and backhaul capacity are considered. Maximizing the total network throughput or the fairness among users in HetNet are challenging mixed integer nonconvex optimization problems. Iterative algorithms based on alternating descent and successive convex programming are proposed to address such problems. Next, we consider a full-duplex multi-user multiple-input multiple-output (FD MU-MIMO) multicell network in which base stations simultaneously serve both downlink (DL) users and uplink (UL) users on the same frequency band via multiple antennas to potentially double the spectral efficiency. Since the use of FD radios introduces additional self-interference (SI) and cross interference of UL between DL transmissions, the minimum cell throughput maximization and the sum network throughput maximization with QoS guarantee are nonconvex challenging problems. To solve such challenging optimization problems, we develop path-following algorithms based on successive convex quadratic programming framework. As a byproduct, the proposed algorithms can be extended to the optimal precoding matrix design in a half-duplex MU-MIMO multicell network with the Han-Kobayashi transmission strategy. Finally, the last research work stems from the need of prolonging user equipments’ battery life in power-limited networks. Toward this end, we consider the optimal design of precoding matrices in the emerging energy-harvesting-enabled (EH-enabled) MU-MIMO networks in which BSs can transfer information and energy to UEs on the same channel using either power splitting (PS) or time switching (TS) mechanisms. The total network throughput maximization problem under QoS constraints and EH constraints with either PS or TS in FD networks is computationally difficult due to nonconcave objective function and nonconvex constraints. We propose new inner approximations of these problems based on which a successive convex programming framework is applied to address them.
Please use this identifier to cite or link to this item: