Channel Estimation for FDD Massive MIMO With Complex Residual Denoising Network

In order to improve the performance of downlink channel estimation in frequency division duplexing (FDD) massive multiple-input multiple-output (MIMO) systems, we propose a channel estimation method based on compressed sensing (CS) and deep learning (DL). To eliminate signal noise, we employ a complex residual denoising network called CR-DnCNN on top of CS processing, which can greatly reduce the computational cost of the network by combining complex convolution and fewer residual blocks. Simulation results show that it can significantly reduce the pilot overhead compared with the classical algorithm and has lower MMSE and BER in the channel estimation process.