Study On Semi-blind Channel Estimation Methods Of MIMO-OFDM System

Multiple-input-multiple-output (MIMO) antenna architecture has the ability to increase capacity and reliability of a wireless communication system. Orthogonal frequency division multiplexing (OFDM) is well-known for efficient high speed transmission and robustness to frequency selective channels. Hence, the integration of the two technologies has the potential to meet the ever growing demands of future communication systems. A major challenge to MIMO-OFDM systems is how to obtain the channel state information accurately and promptly for coherent detection of information sambols and channel synchronization. therefore, channel estimation play an important role in improving the performance of communication systems. The channel estimation methods of MIMO-OFDM system can be divided into non-blind estimation,blind estimation and semi-blind estimation.Non-blind channel estimation methods are achieved by inserting the pilot symbols in the transmit signals, it has the advantage of simple implementation and low complexity, however, due to the use of the pilot, the efficiency of data transmission is very low and these methods can’t track the change of the channel timely. The blind estimation method has the advantage of good performance and high data transmission efficiency, but its implementation complexity is very high and it has a low convergence speed, so it is very difficult to apply these methods in the practical communication systems. The semi-blind channel estimation method combine the advantages of the non-blind and blind, it improve the estimation performance on the basis of reducing the complexity of the algorithm, so it becomes the hot spot in recent years. In this article, we will do some study about the semi-blind channel estimation of MIMO-OFDM system.In the method of semi-blind channel estimation of MIMO-OFDM system, this article focuses on the two different principle based channel estimation methods:subspace based estimation methods and neural network based estimation methods respectively. In the subspace based method, this article introduce an Toeplitz block matrix based semi-blind channel estimation method, the semi-blind channel estimation is investigated by combining few pilot sequences with the subspace method, it can eliminate the ambiguity of the blind channel estimation with the assistance of few pilot sequences, the semi-blind method can get a good estimation performance compared with the blind subspace based estimation method. In the neural network based method, a semi-blind channel estimation technique based on extreme learning machine for MIMO-OFDM systems is proposed. In this method, extreme learning machine is taken as a channel estimator to estimate the channel impulse response. Simultaion results have shown that the ELM estimator performs a better estimation performance and lower computation compare with the other traditional feed-forward neural network based estimator.