With the rapid development of wireless communications, there has been a massive growth in the number of wireless communications users and progressively more new high data rate wireless services will emerge. With these developments taking place, wireless spectral resources are becoming much more scarce and precious. As a result, research on spectrally efficient transmission techniques for current and future communication networks attracts considerable interest. As a promising multi-antenna communication technique, transmit beamforming is widely recognized as being able to improve the capacity of wireless systems without requiring additional spectral resources. In conventional (rank-one) beamforming, each user is served by a single beamformer. For certain transmit beamforming applications, the beamforming performance may be poor if the degrees of freedom in the conventional beamformer design become insufficient.

The scope of this thesis is to address the beamforming performance degradation problems induced by the insufficient degrees of freedom in the beamformer design in certain practical scenarios. In this thesis, a fundamentally new idea of higher-rank (>1) transmit beamforming is proposed to improve the beamforming performance. Instead of a single beamformer assigned to each user, multiple beamformers are designed and correspondingly the degrees of freedom in the beamformer design are multiplied, i.e., the increase of the degrees of freedom consists in the increase of the number of design variables. To implement higher-rank beamforming, the central idea is to combine beamforming with different space time block coding (STBC) techniques. Conventionally, STBCs are used to exploit the transmit diversity resulting from the independent fading for different transmit antennas. However, the use of STBCs in the higher-rank beamforming approaches is not for the sake of transmit diversity, but for the sake of design diversity in the sense of degrees of freedom in the beamformer design.

The single-group multicast beamforming problem of broadcasting the same information to all users is firstly considered in the thesis. It is assumed that the transmitter knows the instantaneous channel state information (CSI) which describes the short-term channel conditions of a communication link and can be estimated in modern communication systems. In the conventional approach, a single beamforming weight vector is designed to steer the common information to all users. In the case of a large number of users, the performance of the conventional approach usually degrades severely due to the limited degrees of freedom offered by a single beamformer. In order to mitigate this drawback, a rank-two beamforming approach is proposed in which two independent beamforming weight vectors are designed. In the rank-two beamforming approach, single-group multicast beamforming is combined with the two dimentional Alamouti STBC, and each user is simultaneously served with two Alamouti coded symbols from two beamformers. The degrees of freedom in the beamformer design are doubled and significant performance improvement is achieved.

The multi-group multicast beamforming problem of transmitting the same information to users in the same group while transmitting independent information to users in different groups, is studied next in the thesis, also assuming that instantaneous CSI is available at the transmitter. The rank-two beamforming approach, originally devised for single-group multicasting networks that are free of multiuser interference, is extended to multi-group multicasting networks, where multiuser interference represents a major challenge. By combining multi-group multicast beamforming with Alamouti STBC, two independent beamforming weight vectors are assigned to each user and the degrees of freedom in the beamformer design are doubled resulting in drastically improved beamforming performance.

Then, the multiuser downlink beamforming problem of delivering independent information to different users with additional shaping constraints is investigated in the thesis, also assuming instantaneous CSI at the transmitter. Additional shaping constraints are used to incorporate a variety of requirements in diverse applications. When the number of shaping constraints is large, the degrees of freedom in the beamformer design can be rather deficient. In order to address this problem, a general rank beamforming approach is proposed in which multiuser downlink beamforming is combined with high dimensional (>2) real-valued orthogonal space time block coding (OSTBC). In the general rank beamforming approach, the number of beamforming weight vectors for each user and the associated degrees of freedom in the beamformer design are multiplied by up to eight times, which lead to significantly increased flexibility for the beamformer design.

Since instantaneous CSI can be difficult to acquire in certain scenarios, the use of statistical CSI describing the long-term statistical characteristics of the channel can be more practical in these scenarios. The rank-two beamformer designs based on instantaneous CSI can be straightforwardly applied in the case of statistical CSI. However, it is impossible to extend the general rank beamforming approach for the multiuser downlink beamforming problem with additional shaping constraints based on instantaneous CSI to the case of statistical CSI straightforwardly. Therefore, multiuser downlink beamforming with additional shaping constraints using statistical CSI at the transmitter is then studied and an alternative general rank beamforming approach is proposed in the thesis. In the general rank beamforming approach using statistical CSI, multiuser downlink beamforming is combined with quasi-orthogonal space time block coding (QOSTBC). The increased number of beamforming weight vectors and the associated degrees of freedom are much beyond the limits that can be achieved by Alamouti STBC in the beamformer design.

Simulation results demonstrate that the proposed higher-rank transmit beamforming approaches can achieve significantly improved performance as compared to the existing approaches.