MIMO wireless communications in frequency selective fading channels

Multiple-input multiple-output (MIMO) wireless communication system exploits the multiple spatial channels to achieve higher data rate or improved performance in multipath fading environment. Various MIMO systems have been proposed recently, where the most well developed and promising schemes are the Vertically-layered Bell Laboratories Layered Space-Time (V-BLAST) system, the Space-Time Trellis Code (STTC), and the Space-Time Block Code (STBC). These schemes assumed flat fading channels, and exploration into frequency selective fading channels is a challenging and necessary footstep for practical high-speed realization. In this thesis, we investigate these MIMO systems in frequency selective fading channels, and propose several schemes using maximum likelihood sequence estimator (MLSE) to combat this performance limiting impairment. The V-BLAST system in these channels is investigated first and three layered detection techniques are proposed, namely the Zero-Forcing Maximal Ratio Combining (ZF-MRC), the Layered Maximum Likelihood Detection (L-MLD), and the Group Maximum Likelihood Detection (G-MLD). In addition, we propose the imaginary transmit antenna idea to facilitate the derivation of these schemes, and the imaginary receive antenna idea to reduce the number of receiving antennas. Analysis showed that the L-MLD and G-MLD schemes outperform other existing solutions. In our second contribution, we analyze the STTC with ML equalization and detection (MLED), and with the orthogonal frequency division multiplexing (OFDM) over frequency selective fading channels. Their performance bounds are derived and their code design criterions, receiver complexity and other issues are compared and discussed. From this analysis, we found that both schemes have different advantages over each other and in particular, the STTC MLED is more preferable in short delay spread channels. However, the STTC MLED assumes perfect channel knowledge in the receiver, but channel estimation is required in practice. To provide reliable channel estimates and reduce unnecessary computations in conventional receivers, we propose a novel Expectation-Maximization (EM) receiver that performs channel gain, channel order, and sequence estimation. Simulation results showed that the proposed scheme outperforms the conventional receiver with lesser complexity, and is highly robust in various channel environments. Finally, we study the concatenation of trellis-coded modulation with Time-Reversal STBC (TR-STBC) for frequency selective fading channels, and obtain a sub-optimal code design criterion. From the simulation results, the concatenated system with the proposed codes achieves better performance with lesser complexity than STTC over the same channel environment