Research On Blind Equalization Algorithms For MIMO System

In Multiple Input and Multiple Output multiplexing system, signal transmitting through the multipath channel, the signal received from each antenna is a mixture of all the transmission signals and multipath signals, which results in the inter-symbol interference(ISI), interchannel interference(ICI). These interferences seriously affect the performance of the system. Without the need of sending training sequences, MIMO blind equalization improves the spectrum utilization rate, and effectively reduce or eliminate the impact of ISI and ICI. Therefore, the study on MIMO blind equalization algorithm is of great significance.In this thesis, different MIMO blind equalization algorithms are studied. The main work is summarized as follows:1. For MIMO Multi-Stage interference cancellation blind equalization model, the reliable recovery of the first source signal affects cancellation of the inter-channel interference. Accordingly, the efficient equalization algorithms for the first stage source signal are proposed, including a single-mode MCMA-MSQD, dual-mode MMCMA-MCME algorithm. At the same time, a channel estimation method is proposed. On the basis of the reliable recovery of the first stage source signal, the later stage source signal will gain better equalization performance through ICI cancellation. Simulation results show that, compared with the MCMA algorithm, the steady-state error of MCMA-MSQD algorithm is decreased by 5d B; Convergence speed of MMCMA-MCME algorithm is faster by 700 symbols and steady-state error is decreased by 5d B.2. Owing to the need to recover source signal one by one for MIMO Multi-Stage blind equalization algorithm, two blind equalization algorithms,which are MCMA-MUK algorithm, combined with blind source separation are proposed to recover source signal simultaneously. In the same conditions, compared to blind equalization algorithm with related items, the first source signal convergence speed of MCMA-MUK algorithms is greatly improved; MCMA-CFPA algorithm has faster convergence speed and less steadystate error than that of MCMA-MUK algorithm. The improved MCMA-CFPA algorithm further reduces the complexity of the algorithm on the premise of equalization performance unchanged.3. Depending on different initialization of equalizer vector, selecting the optimal equalization algorithm is proposed. The algorithm can select initialization with the better equalization performance. In addition, Selective combining blind equalization algorithm is proposed. Compared with selecting the optimal algorithm, selective combining algorithm can obtain the multi-antenna array gain and diversity gain. Simulations show the algorithm can reduce symbol error rate by one to two orders of magnitude. The more the number of receiving antennas, the lower the symbol error rate.