| Multiantenna systems play an important role in wireless communications in terms of improving system throughput and reliability by exploiting the spatial dimension in addition to the traditional dimension of time and frequency. This thesis presents three multiantenna transceiver designs for different applications. The general goal of these designs is to achieve optimized performance by exploiting various statistical properties of the channel in time, space, and polarization.;The integrated VSM-DLA system in Chapter 3 is motivated by the observation that communication over Virtual Spatial Modes (VSM) using the discrete Fourier transform can be realized in an analog fashion by the hardware architecture of a Discrete Lens Array (DLA). Analog front-end processing can result in reduced power consumption compared to digital processing. Another important feature of the system is the use of co-located antennas with orthogonal polarizations to enhance channel capacity without increasing the array dimension. A channel model is proposed to investigate the corresponding capacity improvement as well as the impact of hardware non-idealities.;The adaptive spatial multiplexing scheme in Chapter 4 is motivated by the limitation of the widely used spatially uncorrelated block fading channel model, which ignores temporal channel variation in each packet as well as spatial correlation. Both factors are considered in the proposed design to improve performance. Spatial correlation is exploited by transmitting data streams from different transmit angles with power and rate allocated according to the channel power in those angles. The packet length is further optimized based on temporal channel correlation to maximize the average throughput. The results show that the proposed design based on statistical channel knowledge at transmitter can yield competitive performance compared to the case assuming instantaneous channel state information at transmitter.;In Chapter 5, channel correlation in time and space is jointly considered in the proposed Quality-of-Service (QoS) based space-frequency prefiltering scheme for MultiCarrier-Code-Division-Multiple-Access (MC-CDMA) downlink communications. Temporal channel correlation is exploited to predict the channel over the packet duration from the estimate obtained from the training symbols. Moreover, the results indicate a maximum packet length, beyond which the QoS cannot be ensured due to the channel prediction error. |
