Deep learning-based CSI feedback for IoT-oriented massive MIMO systems

In Internet of Things (IoT)-oriented massive Multiple Input Multiple Output (MIMO) systems, the base station (BS) needs to be accurately informed of the channel state information (CSI) fed by the user equipment (UE). The traditional CSI feedback scheme is difficult to guarantee the communication performance when the number of antennas at the BS increases significantly in massive MIMO systems. Inspired by deep learning (DL) and compressed sensing (CS) algorithms, this paper proposes a CSI feedback scheme CQ-CsiNet. On the basis of CSI dimension compression and reconstruction, the CSI compression codeword representation is converted into CSI bitstream to better meet the needs of the actual communication system. The experimental results show that the feedback scheme proposed in this paper has good CSI feedback accuracy and robust performance in different scenarios, and can effectively compress and reconstruct feedback CSI under different settings of feedback bits and quantization bits, further witnessing the DL great potential in the field of intelligent communication.