Receiver processing and limited-feedback user scheduling for multiuser MIMO and MIMO-OFDM downlink.

摘要

Multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) has been proposed for many emerging standards and seems to be a promising solution for future high data rate wireless communications. In the first part of this thesis, a novel sub-optimum detection method called unified successive interference cancellation (U-SIC) for spatially multiplexed multicarrier code division multiplexing (SM-MC-CDM) transmission is proposed. It is shown that compared to the spatially multiplexed OFDM (SM-OFDM), the frequency domain spreading in SM-MC-CDM systems results in an additional diversity gain. Analytical results for the performance and capacity of zero-forcing (ZF) U-SIC are provided. The results demonstrate significant performance improvement over other existing methods of comparable complexity. Performance of turbo-coded SM-MC-CDM transmission is also investigated.;Next, net throughput is used as a benchmark to compare several MIMO-OFDM downlink transmission schemes with complete CSIT and also with limited feedback. Considering limited feedback per-chunk (a chunk consists of a number of adjacent subcarriers in frequency and consecutive OFDM symbols in time) user scheduling for MIMO-OFDM downlink, it is shown that there exists a chunk size which maximizes the average net throughput. It is shown that the net throughput maximizing chunk size depends on the number of users in the system and the communication channel’s characteristics. Finally, future directions for possible research are given.;In the next part of the thesis, multiuser MIMO downlink is considered. Efficient transmission schemes based on zero-forcing (ZF) linear receiver processing, eigenmode transmission and partial channel state information are proposed. The proposed schemes utilize a handshaking procedure between the BS and the users to select (schedule) a subset of users and determine the precoding matrix at the base station (BS). The advantage of the proposed limited feedback schemes lies in their relatively low complexity scheduling algorithms and high sum rate throughput, even for a small pool of users.