Learning the CSI Recovery in FDD Systems

We propose an innovative machine learning-based technique to address the problem of channel acquisition at base station in frequency division duplex systems. In this context, reconstructs full state information downlink range based on limited feedback from mobile terminal. The recovery is a convolutional neural network which trained exclusively collected samples acquired uplink domain. No training required all. Finally, after detailed presentation and analysis proposed its performance, “reciprocity distribution” assumption that central approach validated with maximum mean discrepancy metric.