| After a few decades' evolution of wireless communication systems, to ensure reliable high-speed communication over unreliable wireless channels is still one of the major challenges facing researchers and engineers. The use of multiple antennas at transmitter and receiver, known as multiple-input multiple-output (MIMO) communications, is one promising technology delivering desired wireless services. The main goal of this thesis is to study two important issues in wireless MIMO communication systems: receiver design for coded MIMO systems, and diversity-multiplexing tradeoff analysis in general fading channels.;In the first part of this thesis, we decompose the receiver design problem into two sub-problems: MIMO channel estimation and MIMO detection. For the MIMO channel estimation, we develop an expectation-maximization (EM) based semi-blind channel and noise covariance matrix estimation algorithm for space-time coding systems under spatially correlated noise. By incorporating the proposed channel estimator into the iterative receiver structure, both the channel estimation and the error-control decoding are improved significantly. We also derive the modified Cramér-Rao bounds (MCRB) for the unknown parameters as the channel estimation performance metric, and demonstrate that the proposed channel estimation algorithm can achieve the MCRB after several iterations. For the MIMO detection, we propose a novel low-complexity MIMO detection algorithm, which has only cubic order computational complexity, but with near-optimal performance. For a 4×4 turbo-coded system, we show that the proposed detector had the same performance as the maximum a posteriori (MAP) detector for BPSK modulation, and 0.1 dB advantage over the approximated MAP detector (list sphere decoding algorithm) for 16-QAM modulation at BER = 10-4.;In the second part of this thesis, we derive the optimal diversity-multiplexing tradeoff for general MIMO fading channels, which include different fading types as special cases. We show that for a MIMO system with long coherence time, the optimal diversity-multiplexing tradeoff is also a piecewise linear function, and only the first segment is affected by different fading types. We proved that under certain full-rank assumptions spatial correlation has no effect on the optimal tradeoff. We also argued that non-zero channel means in general are not beneficial for multiplexing-diversity tradeoff. |
