A Channel Information Compression Feedback Method Based on Deep Recurrent Neural Network

Abstract

The invention discloses a channel information compression feedback method based on a deep cyclic neural network. First, the channel matrix is ​​transformed into the angle-time delay domain through discrete Fourier transform, and then the columns containing non-zero elements in the channel matrix are reserved, and the The complex number matrix composed of these columns is split into a two-channel real number matrix. The user side first uses the convolutional network to extract the frequency domain features of the channel, and then compresses the extracted frequency domain features through the fully connected neural network and the long and short time memory network. The signal is transmitted to the base station through the system feedback link. The base station uses a fully connected neural network and a long-short-term memory network to decompress the signal, and uses a multi-layer convolutional neural network to restore and reconstruct the channel matrix. The invention effectively compresses the time redundancy of the channel by introducing the cyclic neural network, and improves the compression rate of the channel information under the premise of ensuring higher performance.

Description

technical field [0001] The invention belongs to the technical field of compressed sensing and channel information recovery, and relates to a channel information compression feedback method, in particular to a channel information compression feedback method based on a deep cycle neural network. Background technique [0002] Millimeter-wave multiple-input multiple-output (MIMO) is one of the key technologies of the fifth-generation mobile communication, which can ensure that the communication system has ultra-high system capacity, energy efficiency and anti-interference ability. [0003] In millimeter-wave MIMO, the number of antennas on the base station side is usually very large, which leads to downlink pilot information and uplink channel state information (CSI) feedback occupying a large amount of resources. Compressed sensing technology is based on the assumption of channel sparsity, which can compress channel information to a greater extent and restore the original chann...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem that the existing method does not consider time redundancy in the channel compression feedback problem and the feedback accuracy is low, the present invention proposes a channel information compression feedback method based on a deep recurrent neural network, introduces a recurrent neural network, and utilizes a recurrent neural network The memory characteristics of the network better compress the redundancy in timing

Images