Summary: | 碩士 === 輔仁大學 === 電機工程學系 === 100 === Multiple-input multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) has received much attention due to its large potential capacity and high-speed data rates. In wireless communication systems, intersymbol interference (ISI) and intercarrier interference (ICI) are inevitable problems. Therefore, the channel estimation is indispensable to achieve coherent demodulation. The subspace-based blind channel estimation method can easily do the channel identification. However, it suffers from some severe problems in practice. First, due to the property of the subspace-based blind channel estimation, it must receive a large number of the received signals for the sake of estimating the channel. Nevertheless, it is not suitable for the time-variant system in practice. The proposed block matrix scheme is here to solve the problem above. This way will not only enhance the performance of the subspace-based channel estimation but also can use few OFDM blocks to estimate the channel. The block subspace-based channel estimation will achieve faster convergence speed than the conventional one.
In this thesis, we discuss the system all in finite sample scenario and we will focus on our proposed channel estimation. Once we know the information of channel impulse response, we exploit the minimum mean square error (MMSE) equalizer and zero-forcing (ZF) equalizer for symbol detection. Our proposed channel estimation can have obviously lower BER and faster convergence rate in symbol detection, in which illustrate in computer simulations.
|