Abstract：In this paper a novel adaptive semi-blind beamforming algorithm is proposed for OFDM-MIMO systems. First the nonlinear neural networks combined with maximized mutual information rule are used for estimating the coarse weights of beamformer, then the beamforming algorithm adaptively adjusts the vector weights by means of the known pilot tones inserted into the frame structure, thus the channel equalization can be further im-proved. The proposed algorithm is compared to two other channel estimators under various multipath fading environments through simulations and experimental data. The results show that the semi-blind beamforming algorithm has much better overall performance in terms of BER, moreover has high aperture gains with very good directivity in the beamformer pattern.
2004, Vol. 26 
