| In this thesis, we consider the Multiple-Input Single-Output (MISO) downlink multi-user multicast channel, where a group of single-antenna users is to receive the same data stream simultaneously from an N-antenna transmitter. In this context, various system settings are studied, i.e., perfect versus imperfect channel state information at the transmitter side (CSIT), and with ideal Gaussian inputs versus finite-alphabet inputs. For all the scenarios, a popular approach is to apply transmit beamforming, whereby the associated beamforming optimization is handled by a rank-one approximation method called semidefinite relaxation (SDR). The SDR-based beamforming paradigm has been shown to be promising when the number of users served is small, but will experience an obvious performance degradation when the number of user served is large. Therefore, our main interest in this thesis is to derive physically realizable transceiver designs which can provide all users with good information rate simultaneously, irrespective of any factors, e.g., the number of users served in the multicast system.;We begin our work under the system setting with perfect CSIT and Gaussian signal inputs. In this context, we propose a stochastic beamforming (SBF) strategy which randomizes the beamformer in a per-symbol time-varying manner. Specifically, we propose three efficiently realizable transmit schemes and show that they can achieve at worst a constant rate gap of 0.8314 bits/s/Hz with respect to the MISO multicast capacity. This result is significant since, excelling beyond the limitations of prior works, we manage to devise practical transmit schemes that can exhibit the same scaling (with respect to the number of users) as the MISO multicast capacity in general cases, in terms of the multicast rate. As the second contribution of this thesis, we further propose two more strategies. The first strategy combines transmit beamforming and Alamouti space-time coding (STC). The resulting beamformed (BF) Alamouti scheme is a rank-two generalization of transmit beamforming, and is capable of providing better multicast rate than conventional transmit beamforming. The second strategy combines SBF and BF Alamouti to result in an SBF Alamouti scheme. We prove that this scheme can achieve an improved multicast rate gap of 0.39 bits/s/Hz, which is quite close to the multicast capacity. As the third contribution, we consider robust transmit designs in the scenario when the channel state information is imperfect at the transmitter side and the channel inputs are ideal Gaussian. In particular, we consider the worst-case design and the chance-constrained design. For both the design paradigms, via some sophisticated methods, the transmit optimization can be formulated as solving semidefinite programme (SDP) problems. Moreover, we show that the newly proposed transmit schemes can be also applied and provide relatively good rate performance. In particular, we establish the analytical results on the approximation quality for the SDR-based transmit beamforming scheme and BF Alamouti scheme for robust multicasting, which was not addressed in the literature. As the forth contribution, we investigate the MISO multi-user multicast achievable rate maximization with finite-alphabet inputs, given perfect CSIT. We show that, the rate maximization problem with finite-alphabet inputs is actually equivalent to that with ideal Gaussian inputs, which has been well studied in the literature. Therefore, the SDR-based beamforming and BF Alamouti schemes are promising for a small number of users. Nevertheless, for large number of users, the SBF schemes and SBF Alamouti schemes are more desirable in terms of the multicast rate. Finally, extending upon the aforementioned results, we adopt the newly proposed transmit schemes to the scenario with imperfect CSIT and finite-alphabet inputs. This completes our study on all the system settings. |
