Research On Channel Estimation And Coding Algorithms In MIMO-OFDM System

In recent years, the demand for reliable high data rate multimedia and data transmission in wireless communications has increased tremendously. OFDM and MIMO communications provide both theoretically attractive and technically practical solutions to satisfy these requirements.During the graduate period, the author has the great opportunity to participate in the EU research project MAGNET which aims at the standardization of the B3G/4G mobile communications. The author is in charge of the research on SISO-OFDM & MIMO-OFDM channel estimation and MIMO-OFDM space-time coding algorithms under the confinements of the air interface of MAGNET system's physical layer.As for the SISO-OFDM channel estimation algorithms, the existing DFT method can not meet the mean square error performance requirement due to the particular pilots pattern of MAGNET system. In order to alleviate the impact of the spectrum leakage which is caused by the guard band of the OFDM symbol, the author proposes the windowed DFT method which can effectively improve the mean square error performance. As for MIMO-OFDM channel estimation algorithms, the author not only improves and extend the existing time-domain channel estimation method based on MMSE principle, but also derives the optimal preamble design principle from the MMSE aspect.Further the author goes on exploring on the pre-coding of MIMO system which is a very attractive and meaningful research area due to its advantage to improve the throughput and BER performance of the MIMO system significantly when the channel state information could be available at the transmitter in reality. The author extends the order statistic theory into the MIMO system and proposes the algorithm on selection of antennas and modulation order under the constant bits transmitted confinement. This algorithm can significantly improve the BER and the throughput performance and on the contrary to the other bit-loading algorithms, it does not require non-uniform bit and power level allocation among antennas, thus drastically reducing the hardware complexity. The author also gives a general framework under which any principles can be used, such as MMSE, throughput maximization, MBER and so on to make this complicated problem more clearly.All the Simulation results are given in this thesis to show the algorithms that we adopted are very acceptable.