Towards increasing the capacity in MIMO wireless communication systems

The use of multiple antennas on both sides of a multipath channel can improve the capacity without additional transmit power. It is possible, in principle, for the capacity to increase linearly with the number of antennas even when the channel state information (CSI) is unknown at the transmitter. These known capacity results require many assumptions, including the need for the CSI to be known perfectly. In practice, perfect CSI is never known perfectly a priori, and its estimation, without using an ideal blind technique (none are available), requires bandwidth resource which reduces the capacity. Moreover, various factors such as digital modulation, finite block lengths, and imperfect power allocation degrade the capacity from the Shannon limit to the practicable possibilities of a digital link. These practical impairments motivate new techniques for increasing the capacity in MIMO systems, and in this thesis, two sets of techniques are presented. The first set includes two new decision-directed techniques for estimating the channel matrix, and they are shown to have higher capacity compared to pilot symbol assisted modulation systems since no pilots are required. These techniques are applicable for various open-loop SISO/MIMO wireless communications systems including systems employing OFDM, nonlinear/linear equalization, MRC, Alamouti coding, and spatial multiplexing. In the second set, the eigen-MIMO capacity is maximized in the presence of different practical impairments. In particular, the joint influence of training-based channel estimation and imperfect feedback on both the information-theoretic and the practicable water-filled eigen-MIMO capacities, are analyzed. Water-filled eigenchannels maximize the information theoretic capacity, but for implementation, the required adaptive modulation means high complexity. One simplification is to have fixed modulation over a fixed number of eigenchannels. However, the error rate deteriorates with the weakest eigenchannel and to counter this while maintaining high throughput, the information rate is maximized with an output SNR constraint. On the other hand, if higher complexity can be tolerated, adaptive modulation and coding can be deployed for high throughput. In this context, a high capacity eigen-MIMO system using Reed-Solomon coded M-QAM is developed. This includes an appropriate QAM, code rate, and power allocation for each eigenchannel.