Channel state information feedback and reconstruction algorithm for massive MIMO based on joint attention mechanism neural networks

This paper studies the feedback and reconstruction of Channel State Information (CSI) in the downlink of FDD mode of massive multiple input multiple output (MIMO) system. Aiming at the problems such as high computational complexity and strict requirement on channel sparsity of traditional CSI feedback algorithms based on compressed sensing, and the problem that existing CSI feedback algorithms based on deep learning are insufficient in performance. A massive MIMO channel state information feedback and reconstruction algorithm based on joint attention mechanism neural network is proposed. And it includes a new CA-CsiNet neural network model based on auto-encoder neural network architecture. The attention mechanism algorithm is introduced into the encoder network to improve the feature extraction ability of the neural network effectively. Experimental results show that the proposed algorithm has better performance than traditional CSI feedback and reconstruction algorithms and existing deep learning algorithms.