Design And Implementation Of Semi-Blind Channel Estimation Based On Matrix Decomposition For MIMO-OFDM Systems

During the third wireless communications, using multiple-input and multiple-output technology, MIMO system can increase the channel capacity multiplied without increasing the average power and the bandwidth of the system. Compared with the signal-input and signal-output system, MIMO system has a higher spectrum efficiency with the same bandwidth and total signal power. OFDM technique can reduce the side-effect factor of the frequency selective channel environment using the orthogonality of the sub-carriers.In order to exploit diversity and implement equalization and decoding in the receiver, accurate channel state information is necessary. Therefore, efficient channel estimation methods for MIMO-OFDM systems is a hot topic for a long period.In order to accomplish the channel estimation for MIMO-OFDM system, three methods have been proposed. The first method is the pilot-added channel estimation using the pilots or training sequences inserted into the transmitting signals. The second method is the blind channel estimation using the statistic characteristics of received signals only. The last method is the combination of the pilot-added and blind methods. In this method, pilots or training sequences and the statistic characteristics of received signals are both used.In this paper, the background and meaning of channel estimation methods for MIMO-OFDM are first introduced including the principle of the three channel estimation methods and the realistic meaning of channel estimation for MIMO-OFDM system. In the following, the principle of the OFDM technique and the MIMO-OFDM system model are discussed. The main research of this paper is focused on the model of multi-user and multi-antenna OFDM system, which is based on MIMO-OFDM system. Under this model, two semi-blind channel estimation methods are proposed, that are the improved WR decomposition method and the improved NLMS based method. Computer simulation results show that this improved WR decomposition method has a better performance in tracing the channel responses compared with the normal LS and MMSE channel estimation methods. In the end under the same model, adaptive channel estimation methods are discussed. Computer simulation results and theoretical analysis show that this improved NLMS based method has a better performance in increasing the convergence speed and reducing the computing complexity.