Semi-Blind Channel Estimation Methods Of MIMO-OFDM System Based On Neural Network

The combination of orthogonal frequency-division multiplexing (OFDM) and multiple-input multiple-output (MIMO) technologies, which referred to as MIMO-OFDM, is currently under study as one of the most promising candidate for next-generation communications systems, ranging from wireless LAN to broadband access. Recent works have tackled the performance assessment (through both simulation and measurements) of MIMO-OFDM systems, such as the problem of channel estimation.Channel estimation is the key part of the design of any receiver, the number of channel in MIMO-OFDM system will be increased with the growing of antennas and users'number rapidly, and the channel estimation of MIMO-OFDM system is a very complex question. Although a large number of studies are available on the study of channel estimation, but many of them are only applicable to flat fading channels. Common channel estimation method is divided into two categories:one is known by additional information (training sequence or pilot), training method is very simple but needs to consume some bandwidth, and bandwidth is a very precious resource in wireless communications system; the other one is blind estimation methods, which do not consume bandwidth, but have more complex to achieve. So we can use a small pilot and the iterative method, which called semi-blind estimation, to reduce the complexity of estimation and improve the channel estimation in real time to a certain extent.In nonlinear MIMO-OFDM system, it is necessary for the intelligent signal processing method applied to deal with channel estimation. Neural network can be a good model of the nonlinear system. The channel estimation method that based on neural network can reduce the computational burden, and has better convergence rate and estimation accuracy. So we proposed a semi-blind estimation methods based on the RBF network and Hopfield neural network, then combination of ICA algorithm and Hopfield neural network algorithm for MIMO-OFDM system channel estimation. Finally, the simulation results show that the advantages of neural network.Finally, this article summarizes the work, and points out the aspects that need continue to improve and research.