Channel Modelling and Estimation in Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing Wireless Communication Systems

In wireless communications, the demands for high data rates, enhanced mobility, improved coverage, and link reliability have enormously increased in recent years and are expected to further increase in the near future. To meet these requirements, new concepts and technologies are needed. Theoretical studies have shown that using multiple antennas at the transmitter and receiver, known as multiple-input multipleoutput (MIMO) technology, can dramatically increase the capacity, coverage, and link reliability of a communication system. Orthogonal frequency-division multiplexing (OFDM) is an attractive technique for high data rates transmission over frequency-selective fading channels, due to its capability in combating the intersymbol interference (ISI). The combination of MIMO and OFDM results in a powerful technique that incorporates the advantages of both MIMO and OFDM, and is a strong candidate for fourth generation (4G) wireless communication systems. In this thesis, two issues related to realizing practical mobile MIMO OFDM communication systems are addressed. The first issue is about MIMO channel modeling and effect of realistic channels on the theoretical capacity. For this target, a geometrically-based three-dimensional (3-D) scattering MIMO channel model is developed. The correlation expressions are derived and analytically evaluated. The impact of spatial correlation on MIMO channel capacity is investigated under different antenna array configurations, angular energy distributions, and parameters. Analytical and numerical results have shown that the elevation angle has considerable effect on the spatial correlation and consequently on the MIMO channel capacity for the case when the antenna array of the mobile station (MS) is vertically oriented. This has led to a conclusion that 3-D scattering MIMO channel modeling is necessary for accurate prediction of MIMO system performance. The second issue addressed in this thesis is the channel estimation in MIMO OFDM systems. New time-domain (TD) adaptive estimation methods based on recursive least squares (RLS) and normalized least-mean squares (NLMS) algorithms are proposed. These estimators are then extended to blindly track the time-variations of the channel in the decision-directed (DD) mode. Simulation results have shown that TD adaptive channel estimation and tracking in MIMO OFDM systems is very effective in slow to moderate time-varying fading channels. It was observed that the performance of the DD RLS-based estimator always outperform that of the DD NLMS estimator at low mobility and low SNR. In contrast, it was found that the DD NLMS estimator gives better tracking performance at moderate mobility and higher SNR. However, as the training rate is reduced, comparable performance with both estimators is obtained at high SNR. Finally, it has been shown that channel estimation in TD is more accurate with less complexity compared to its counterpart in frequency-domain (FD)