Use this URL to cite or link to this record in EThOS:
Title: A rate-splitting approach to multiple-antenna broadcasting
Author: Joudeh, Hamdi
ISNI:       0000 0004 6059 2001
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Signal processing techniques for multiple-antenna transmission can exploit the spatial dimension of the wireless channel to serve multiple users simultaneously, achieving high spectral efficiencies. Realizing such gains; however, is strongly dependent on the availability of highly accurate and up-to-date Channel State Information at the Transmitter (CSIT). This stems from the necessity to deal with multiuser interference through preprocessing; as receivers cannot coordinate in general. In wireless systems, CSIT is subject to uncertainty due to estimation and quantization errors, delays and mismatches. This thesis proposes optimized preprocessing techniques for broadcasting scenarios where a multi-antenna transmitter communicates with single-antenna receivers under CSIT uncertainties. First, we consider a scenario where the transmitter communicates an independent message to each receiver. The most popular preprocessing techniques in this setup are based on linear precoding (or beamforming). Despite their near-optimum rate performances when highly accurate CSIT is available, we show that such techniques exhibit severe losses under CSIT uncertainties, even when optimally designed. We depart from this conventional approach and adopt an unorthodox transmission strategy based on Rate-Splitting (RS), which relies on broadcasting a common data stream on top of the private data streams precoded using partial CSIT. We propose an average Weighted Minimum Mean Square Error (WMMSE) algorithm to maximize the ergodic sum-rate performance. While the ergodic sum-rate measure captures the long-term overall performance, it is not well suited for delay-limited or fairness based transmissions. Hence, we generalize the RS strategy to formulate the problem of achieving robust max-min fairness over one random fading state under a bounded CSIT uncertainty model. We derive new performance limits in terms of the symmetric-DoF under heterogeneous CSIT qualities across users to identify the RS gains. Then, a robust WMMSE algorithm based on the cutting-set method is proposed to solve the semi-infinite optimization problem. This framework is extended to address the problem of power minimization under Quality of Service (QoS) constraints. Finally, we consider the problem of achieving max-min fairness in a multigroup multicasting scenario, where each message is intended to a group of users. We assume perfect CSIT in this setup, where the presence of multiple users in each group is thought of as a source of (finite) uncertainty. The DoF performance of conventional beamforming techniques are derived from which their limitations are identified. The RS strategy is then extended to this scenario, where we show that significant DoF gains can be achieved. The RS precoder optimization problem in this setup is then solved using the WMMSE approach.
Supervisor: Clerckx, Bruno Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral