Partial channel knowledge based precoding for MIMO and cooperative communications

Linear precoding can be viewed as a convenient way to enhance the throughput and performance of a Multi-Input-Multi-Output (MIMO) communications system for its linear nature that facilitates its low-complexity implementation in a transceiver. Nevertheless, linear precoding designs usually assume instantaneous channel responses perfectly known at the transmitter, which is unrealistically difficult (if not impossible) in fast time-varying channels.;An alternative to the assumption of full channel state knowledge is to consider the availability of partial channel knowledge at the transmitter, including information that changes much slower than the instantaneous channel responses such as the channel statistical parameters. The work presented in this thesis focuses on the design of precoding schemes that are mainly based on partial channel information.;Starting with point-to-point MIMO systems, we show that it is possible to design suitable precoding schemes that achieve a considerable gain in terms of performance or throughput by considering the spatial and path correlation matrices of the frequency-flat and frequency-selective fading channels, respectively.;Next, we investigate the problem of transmission in point-to-multipoint MIMO systems based on partial channel information. We show that in a MIMO broadcast system, using a partial channel knowledge-based user selection scheme in conjunction with precoding can provide an asymptotic optimum sum-rate performance for a growing number of users.;Using the results obtained in the previous steps, we consider the problem of transmission in cooperative relay networks. We first examine and identify the similarities and differences between MIMO and cooperative relay systems especially in term of diversity and multiplexing gain and their trade-off. We then develop the possible solutions for transmission and reception in the relay networks for both cases: single-antenna and multiple-antenna relay nodes. When a large number of single-antenna relay nodes available in the system, we show that a partial channel-knowledge based relay selection scheme in conjunction with a distributed BLAST transmission scheme can achieve the optimum multiplexing-diversity trade-off. In the case of multi-antenna relay, we first derive the optimum combining scheme at the destination for both amplify-and-forward (AF) and decode-and-forward (DF) relay systems, and then address the optimal precoding designs at the source and relay nodes. We show that a generalized maximum ratio combining (GMRC) in conjunction with linear precoding can offer the optimum received SNR for both AF and DF relay systems.