| This thesis addresses the topic of multi-user channel estimation in a wireless system that combines Orthogonal Frequency Division Multiplexing (OFDM) and Adaptive Antenna Arrays (AAA) for increased spectral use and efficiency. Two current techniques for Time-Invariant (TI) channels are reviewed: a Least Squares (LS) estimator and a simplified low complexity estimator. Then, we present three new multi-user channel estimation techniques: the Constrained Inverse Deconvolution (CID) method, Iterative Low Complexity (ILC) method that extrapolates a Matched Filter (MF) output, and a filtering approach to MF output extrapolation based on the Extrapolation Matrix (EM). Simulation results show that the performance of the CID and the ILC methods can approach that of the LS technique in an iterative fashion, and equal it after only few iterations. The EM estimation method involves a simple single matrix multiply to carry out the estimation, such that the performance is equal to that of the LS method. The three new estimation techniques require less computational complexity than that required by the LS method.; We also review a channel interpolator for Time-Varying (TV) channels: the Parameterized Time-Invariant Doppler (PTID) channels interpolator, based on the channel estimates obtained using the channel estimation techniques presented for TI channels. Then, we present two alternate channel interpolation techniques for TV channels based on the Discrete Fourier Transform and the Ideal Filtering principle, respectively. We also improve the PTID interpolation approach to achieve identical performance with a reduction in the computational complexity by a factor equal to at least the number of the simultaneously active SDMA users. |
