Research on smart antenna beamforming by generalized regression neural network

Abstract

Based on array signal processing, the method that smart antennas beamforming by general neural network in wireless communication is proposed, which improves the accuracy of algorithms, such as MVDR algorithms, DOA algorithms etc. 10-elements uniform line array is simulated by computer, Simulation results have shown that we can obtain the extremely approximate weight vector by using the neutral network method. The method is much better than MVDR in computing speed. The proposed method is effective. Computer simulation experimental results are in line with the theoretical ones.