Abstract

Multiple-input multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) techniques have been considered as a strong candidate for the next-generation - wireless communication systems, due to their well-known advantages in high data-rate wireless transmission as well as high frequency spectrum efficiency. In the mean time, channel state information (CSI) is required for precise detection and recovery of signals. Therefore, channel estimation plays a significant role in MIMO-OFDM systems. On the other hand, due to the high mobility of wireless terminals, Doppler shift (DS) can be one of the major side-effects of utilizing MIMO-OFDM techniques, which may lead to severe performance loss. Many schemes on DS estimation have been developed for broadband single-input single-output (SISO) systems. A commonly used method is to exploit the auto-correlation property of the channel impulse response (CIR) estimated by well-developed channel estimation approaches, which not only has high accuracy but also moderate computational complexity. Hence, we first investigate an efficient channel estimation method in this thesis. We will then focus on Jakes' model based DS estimation schemes, and further extend to independently identically distributed (i.i.d.) MIMO-OFDM fading channels with both Rayleigh and Rician distributions. In the first part of the thesis, a training-sequence (TS) based least square (LS) channel estimation scheme is presented for MIMO-OFDM systems along with plenty of computer simulations and corresponding analyses. Experimental study shows that the CIR estimates obtained by the LS method are reliable under moderate channel conditions, and can efficiently be utilized for DS estimation. The second part of the thesis first studies the auto-correlation function (ACF) based DS estimation schemes for SISO-OFOM systems in Rayleigh fading channels, and then extends it to Rician fading channels by developing a new approach along with the analysis of its accuracy and complexity. Thereafter, we apply those approaches to MIMO-OFOM systems and present a few enhanced methods by using non-linear interpolation under certain circumstances. Detailed computer simulations and comparisons are performed, confirming that the proposed ACF based schemes give satisfactory estimation performance over i.i.d. Rayleigh or Rician fading channels with various channel conditions.