Study On Channel Estimation And Signal Detection Technology Of MIMO-OFDM Systems In Fast Time-Varying Channels

The MIMO-OFDM technology has become the mainstream physical layer scheme applied in broadband mobile communication systems, with advantages in resisting frequency selective fading and boosting system capacity. Nevertheless, the rapid channel variance in a high speed system could lead to large Doppler frequency shift, which would bring about serious inter-carrier interference(ICI) at the receiving terminal and system performance deterioration.Here in this paper researches on channel estimation and signal detection at the receiver of MIMO-OFDM system in fast time-varying environment were carried out, which contain:① The characteristics of time and frequency response in fast time-varying environment was analyzed. Linear time-varying model(LTV) and basis expansion model(BEM) were selected to model the time-varying channel. By comparison of the modeling accuracy of different channel simulation models and the impacts that order Q made on BEM, L-BEM with strong orthogonality, high fitting precision and simple structure was selected as the channel model.②Channel estimation algorithm based on LTV model and BEM model for MIMO-OFDM system was studied, where the BEM model showed better adaptability to larger Doppler and better fitting degree. LMMSE estimator based on L-BEM had good performances but required complex information statistics and process. While for LS estimator based on L-BEM, no prior information was needed, and the proliferation of ICI could be suppressed through pre-windowing before LS estimation. On the premise of no obvious increase in algorithm complexity, the performance of the LS estimator was improved. The simulation results demonstrated that this scheme has achieved excellent estimation accuracy and it acquired better performance than LMMSE estimator under the condition of high SNR.③Traditional ordered successive interference cancellation(OSIC) considered single subcarrier signal as the detection layer, and its direct-judging caused serious error propagation. Aiming at this problem, an improved OSIC detection scheme was proposed in this paper. In the coded system, each OFDM symbol was selected as the detection layer, and channel columns norm as the sorting criterion. The convolution was applied to correct the random errors caused by noise, and the interleaving was applied to correct the burst errors caused by depth fading in each layer, to reduce the error propagation and improve the detection precision. The simulation results showed that, compared with traditional detection methods, the new detection method improved the BER performance of the system.④ The serious ICI induced by large Doppler in traditional schemes like LTV-PIC-DSC could bring about irrevocable error propagation, we proposed a windowed BEM-MMSE-OSIC iterative receiving scheme which is based on channel estimation and signal detection technology. The LS estimator based on windowed L-BEM was used to acquire frequency domain channel matrix with strong banded property. And simplified sub matric which only considered the adjacent subcarriers was used to initially eliminate the ICI effect, reducing computing complexity and performance loss. Then the improved OSIC algorithm was used to perform iterative detection, to further eliminate the ICI and IAI. Simulation results proved that the scheme can keep outstanding performance in the case of large Doppler frequency shift.