Study And Realization Of Channel Estimation Techniques For MIMI-OFDM Systems

In order to meet the demands of high spectrum efficiency and high data rate with limited bandwidth in future wireless communaditon, MIMO-OFDM techniques have occurred recently. Combining with the advantage of OFDM and MIMO techniques, MIMO-OFDM technology has been the core technique of next generation wireless communaditon system. In MIMO-OFDM systems, accurate channel state information is necessary at the receiver for coherent demodulation and space-time decoding, so channel estimation techniques are regarded as a key technology and being studied widely. In this thesis, the channel estimation techniques in MIMO-OFDM systems are investigated, and the emphasis is put on training sequence and pilot-based channel estimation methods. Considering the hardness of calculating the inverse matrix and the flaw of DFT algorithm, two improved algorithms are studied respectively.First of all, the principle of OFDM and MIMO technologies are introduced, space-time coding and decoding methods are elaborated. On this basic, the framework of MIMO-OFDM system and its key technologies are descrided.Secondly, we carry out a detailed study on the MIMO-OFDM system channel eatimation. Training sequnence based channel estimation is mainly researched, compared the performance of these algorithm, criteria of LS, MMSE, LMMSE and LMMSE-SVD for channel estimation are atudied. As the difficulty of getting the inverse of matrixs when using teaditional training sequence, a block circulant PN sequence is gave. This new method has good performance, and it is valueable to transform the large matrix inversion problem into a small matrix inversion.Then, the channel estimation methods based on pilot are studied, the interpolation algorithm using comb-type pilot is emphasized. Among the transform domain channel estimation methods, the DFT and DCT interpolation channel estimation methods are focused on, and improved Kalman filter DCT interpolation and FIR filter DCT interpolation algorithm are put forward. The simulation and theoretical analysis shows that the performances of the two algorithms are better than DCT interpolation algorithm.Finally, based on IEEE802.11a architecture, we have realized the LS-based channel estimation algorithm on FPGA. In the simulation system, combination channel estimation method based on training sequence and pilot is utilized, the modulation and demodulation, STBC code and decoding are also included. Verilog simulation results show that it is inaccord with MATLAB simulation results, the correctness of hardware design is proven.