Efficient Wideband Beamforming With Null Broadening Using Fusion Convolutional Block

To meet the development requirements of bandwidth and efficiency in future 6G wireless communications, an efficient wideband beamforming (WB) algorithm is proposed based on fusion convolutional block neural network (FCBNN) in this paper, named as WB-FCBNN algorithm. The presented algorithm can significantly enhance the real-time performance while obtaining a superior output signal to interference plus noise ratio (SINR) of WB. Firstly, the algorithm utilizes null broadening algorithm with interference plus noise covariance matrix (INCM) reconstruction to generate labels, avoiding the influence of interference direction of arrival (DOA) estimation errors. Then, a novel FCBNN framework is constructed to enhance the real-time performance and the output SINR. Specifically, a fusion convolutional block (FCB) is introduced into the FCBNN framework, which can fuse the asymmetric convolution and the standard convolution into one convolution. In this way, the network parameters will be reduced, thus the computational complexity of the network can be reduced. Furthermore, the FCBNN model can be trained via the label. Finally, the well-trained FCBNN model can quickly predict the near-optimal weight vector of WB. Simulation results demonstrate that the presented algorithm can effectively increase the real-time performance of WB, and still has a satisfactory SINR in the environment of high SNR and low snapshot.