Research On Channel Estimation Algorithm For MIMO-OFDM

Orthogonal Frequency Division Multiplexing(OFDM) is a multi-carrier modulation technology, which withstands multipath fading very well, and multiple-input multiple-output (MIMO) system could improve the capacity of the communication system and spectral utilization ratio in case of not increasing bandwidth. As a combination of OFDM with MIMO, MIMO-OFDM has been considered as the core technique of future wireless transmission, because it's the integration of the advantages of MIMO and OFDM. Taking MIMO-OFDM system as the background, both training-sequence-based channel estimation algorithm and pilot-symbols-based channel estimation algorithm are researched intensively in this thesis.Firstly, this thesis presents background and significance of MIMO-OFDM system, characteristics and models of wireless channel. The basic principles and structures of OFDM, MIMO and MIMO-OFDM systems are discussed intensively. Then the channel capacity and space-time coding are analyzed by the MATLAB simulation. The performances of block pilots and comb pilots in different channel conditions are simulated and analyzed.Secondly, the channel estimation algorithms which based on training sequence are researched. According to the shortcomings in time domain channel estimation using random sequence, an optimal training sequence is proposed, which has lower PAPR and needs less computation than random sequence. Towards efficient bandwidth systems without cyclic prefix, the channel estimation algorithms and decoding algorithm applicable are researched. For frequency domain estimation algorithms, a special double training sequences are used to improve the performance, combined with DFT algorithm to reduce noise, then the accuracy of channel estimation is further improved.Finally, the channel estimation algorithms based on pilot symbols are researched. On the basis of channel parameters of pilot subcarriers, linear, Gauss and DFT interpolation algorithms are used to recover the channel parameters. To further improve the performance of DFT interpolation, adaptive filtering algorithms are added to the DFT, four adaptive filtering algorithms are discussed by theoretical analysis and MATLAB simulation. Taking into account the characteristics that the energy of channel time domain impulse response concentrates on some taps, and the computation complexity of adaptive filtering algorithm is rather high, an improved algorithm is researched, which can keep the performance of the original algorithm and reduce the computation.