Method and apparatus for MIMO CSI feedback
A user equipment (UE) reports channel state information (CSI) to a base station in multiple-input multiple-output (MIMO) transmissions. The UE constructs a CSI matrix based on a CSI reference signal (CSI-RS) received from the base station. The CSI matrix is at least three-dimensional in a transmit (Tx) antenna domain, a frequency domain, and a time domain. The UE transforms the CSI matrix into a transformed matrix in at least a Tx beam domain, a delay domain, and a Doppler domain. The UE encodes the transformed CSI matrix into a one-dimensional feature vector using a multi-layered neural network, and sends the one-dimensional feature vector to the base station.
1 . A method for a user equipment (UE) to report channel state information (CSI) to a base station in multiple-input multiple-output (MIMO) transmissions, comprising:
constructing a CSI matrix based on a CSI reference signal (CSI-RS) received from the base station, wherein the CSI matrix is at least three-dimensional in a transmit (Tx) antenna domain, a frequency domain, and a time domain;
transforming the CSI matrix into a transformed matrix in at least a Tx beam domain, a delay domain, and a Doppler domain;
encoding the transformed CSI matrix into a one-dimensional feature vector using a multi-layered neural network; and
sending the one-dimensional feature vector to the base station.
2 . The method of claim 1 , wherein the CSI matrix further includes a fourth dimension in a receive (Rx) antenna domain which is transformed into an Rx beam domain.
3 . The method of claim 2 , wherein the Rx antenna domain is indexed by the number of Rx antennas at the UE.
4 . The method of claim 1 , wherein the time domain is indexed by the number of symbols or slots.
5 . The method of claim 1 , wherein the frequency domain is indexed by the number of subbands or subcarriers.
6 . The method of claim 1 , wherein the Tx antenna domain is indexed by the number of Tx antennas at the base station.
7 . The method of claim 1 , wherein the CSI matrix is a MIMO channel matrix.
8 . The method of claim 1 , wherein the CSI matrix is an orthogonal matrix calculated from a singular value decomposition (SVD) of a MIMO channel matrix.
9 . The method of claim 1 , wherein the CSI matrix is a covariance matrix of a MIMO channel matrix.
10 . The method of claim 1 , wherein the multi-layered neural network includes one or more convolutional neural network layers.
11 . A method for a base station to obtain channel state information (CSI) from a user equipment (UE) in multiple-input multiple-output (MIMO) transmissions, comprising:
receiving a one-dimensional feature vector from the UE;
decoding the one-dimensional feature vector into a matrix of at least three-dimensional in a Tx beam domain, a delay domain, and a Doppler domain using a multi-layered neural network;
transforming the matrix into a CSI matrix in at least a Tx antenna domain, a frequency domain, and a time domain; and
precoding downlink transmission based on the CSI matrix.
12 . The method of claim 11 , wherein the CSI matrix further includes a fourth dimension in a receive (Rx) antenna domain which is transformed from an Rx beam domain.
13 . The method of claim 12 , wherein the Rx antenna domain is indexed by the number of Rx antennas at the UE.
14 . The method of claim 11 , wherein the time domain is indexed by the number of symbols or slots.
15 . The method of claim 11 , wherein the frequency domain is indexed by the number of subbands or subcarriers.
16 . The method of claim 11 , wherein the Tx antenna domain is indexed by the number of Tx antennas at the base station.
17 . The method of claim 11 , wherein the CSI matrix is a MIMO channel matrix.
18 . The method of claim 11 , wherein the CSI matrix is an orthogonal matrix calculated from a singular value decomposition (SVD) of a MIMO channel matrix.
19 . The method of claim 11 , wherein the CSI matrix is a covariance matrix of a MIMO channel matrix.
20 . The method of claim 11 , wherein the multi-layered neural network includes one or more convolutional neural network layers.