Granted Patent B2

US 12,556,243 · App. 18/576,233 · Granted Feb 17, 2026

Inventors: Xueying Hou (Lund, SE); Mats Åhlander (Solna, SE); Sebastian Faxér (Stockholm, SE); Krister Edström (Hjärup, SE)

Assignee: Telefonaktiebolaget LM Ericsson (publ)

H04B7/0456H04L25/021

Abstract

Claims (128)

1 . A method performed by a radio access network, RAN, node, the method comprising:

dividing a channel matrix, H, into two or more sub-matrices, wherein:

the channel matrix, H, is a full channel matrix of one subcarrier of a Multiple-Input-Multiple-Output, MIMO, channel between an antenna array of the RAN node and a particular User Equipment, UE; and

each sub-matrix of the two or more sub-matrices comprises two or more column vectors of the channel matrix, H;

forming a re-ordered channel matrix, H re-order , as a concatenation of the two or more sub-matrices;

forming a port-sorting matrix based on eigen vector matrices, U g i , and eigen value matrices, D g i , obtained via Eigen Value Decompositions, EVDs, performed on channel covariance matrices for the two or more sub-matrices;

applying the port-sorting matrix to the re-ordered channel matrix, H re-order , to provide a port-sorted channel matrix, H re-order ps ;

re-ordering column vectors in the port-sorted channel matrix, H re-order ps , based on eigen values obtained from the EVDs performed on the channel covariance matrices for the two or more sub-matrices to provide a re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ; and

applying grouping of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , and port sorting accordingly to obtain a final port-sorted channel matrix, {tilde over (H)} final ; and

transmitting a downlink transmission to the particular UE based on the final port-sorted channel matrix or the first L columns of the final port-sorted channel matrix.

2 . The method of claim 1 wherein dividing the channel matrix, H, into the two or more sub-matrices comprises:

computing correlation values, ρ i,j , for all pairs of column vectors in the channel matrix, H; and

dividing the channel matrix, H, into the two or more sub-matrices based on the correlation values, ρ i,j .

3 . The method of claim 1 wherein the two or more sub-matrices consist of a first sub-matrix and a second sub-matrix, and dividing the channel matrix, H, into the two or more sub-matrices comprises:

finding the pair of column vectors in the channel matrix, H, having a highest correlation value from among the correlation values, ρ i,j , for all pairs of column vectors in the channel matrix, H;

selecting one column vector from the pair of column vectors in the channel matrix, H, having the highest correlation value as a reference column vector;

selecting the pair of column vectors in the channel matrix, H, having a highest correlation value and N r /2−2 additional channel vectors having the highest correlation values with respect to the reference column vector as the first sub-matrix; and

selecting the N r /2 that are not included in the first-matrix as the second sub-matrix.

4 . The method of claim 1 wherein forming the port-sorting matrix comprises:

for each sub-matrix of the two or more sub-matrices:

computing a channel covariance matrix for the sub-matrix; and

performing an Eigen Value Decomposition, EVD, on the channel covariance matrix for the sub-matrix; and

forming the port-sorting matrix based on the eigen vector matrices, U g i , and the eigen value matrices, D g i , obtained via the EVDs performed on the channel covariance matrices for the two or more sub-matrices.

5 . The method of claim 1 wherein the two or more sub-matrices consist of a first sub-matrix and a second sub-matrix, and the port-sorting sub-matrix is computed as:

U

=

[

U

g

1

*

0

0

U

g

2

*

]

[

D

g

1

-

1

2

0

0

D

g

2

-

1

2

]

wherein U g 1 is the eigen vector matrix for the first sub-matrix, D g 1 is the eigen value matrix for the first sub-matrix, U g 2 is the eigen vector matrix for the second sub-matrix, and D g 2 is the eigen value matrix for the second sub-matrix.

6 . The method of claim 1 wherein applying grouping of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , and port sorting accordingly to obtain the final port-sorted channel matrix, {tilde over (H)} final , comprises:

dividing the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , into two or more sub-matrices each comprising two or more consecutive column vectors from the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

for each sub-matrix of the two or more sub-matrices of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps :

computing a channel covariance matrix for the sub-matrix of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ; and

performing an EVD on the channel covariance matrix for the sub-matrix of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

forming a second port-sorting matrix based on eigen vector matrices, N s k , and the eigen value matrices, L s k , obtained via the EVDs performed on the channel covariance matrices for the two or more sub-matrices of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ; and

applying the second port-sorting matrix to the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , to provide the final port-sorted channel matrix, {tilde over (H)} final .

7 . The method of claim 1 further comprising performing layer selection for the particular UE by selecting a first L columns of the final port-sorted channel matrix, {tilde over (H)} final , wherein L is a number of layers for a current rank of the particular UE.

8 . The method of claim 1 further comprising computing beamforming weights for a downlink transmission to the particular UE based on the final port-sorted channel matrix, {tilde over (H)} final , or the first L columns of the final port-sorted channel matrix, {tilde over (H)} final .

9 . A radio access network, RAN, node comprising processing circuitry configured to cause the RAN node to:

divide a channel matrix, H, into two or more sub-matrices, wherein:

the channel matrix, H, is a full channel matrix of one subcarrier of a Multiple-Input-Multiple-Output, MIMO, channel between an antenna array of the RAN node and a particular User Equipment, UE; and

each sub-matrix of the two or more sub-matrices comprises two or more column vectors of the channel matrix, H;

form a re-ordered channel matrix, H re-order , as a concatenation of the two or more sub-matrices;

form a port-sorting matrix based on eigen vector matrices, U g i , and eigen value matrices, D gi , obtained via Eigen Value Decompositions, EVDs, performed on channel covariance matrices for the two or more sub-matrices;

apply the port-sorting matrix to the re-ordered channel matrix, H re-order , to provide a port-sorted channel matrix, H re-order ps ;

re-order column vectors in the port-sorted channel matrix, H re-order ps , based on eigen values obtained from the EVDs performed on the channel covariance matrices for the two or more sub-matrices to provide a re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ; and

apply grouping of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , and port sorting accordingly to obtain a final port-sorted channel matrix, {tilde over (H)} final ; and

transmitting a downlink transmission to the particular UE based on the final port-sorted channel matrix or the first L columns of the final port-sorted channel matrix.

10 . The RAN node of claim 9 wherein, in order to divide the channel matrix, H, into the two or more sub-matrices, the processing circuitry is further configured to cause the RAN node to:

compute correlation values, β i,j for all pairs of column vectors in the channel matrix, H; and

divide the channel matrix, H, into the two or more sub-matrices based on the correlation values, ρ i,j .

11 . The RAN node of claim 9 wherein the two or more sub-matrices consist of a first sub-matrix and a second sub-matrix, and, in order to divide the channel matrix, H, into the two or more sub-matrices, the processing circuitry is further configured to cause the RAN node to:

find the pair of column vectors in the channel matrix, H, having a highest correlation value from among the correlation values, ρ i,j , for all pairs of column vectors in the channel matrix, H;

select one column vector from the pair of column vectors in the channel matrix, H, having the highest correlation value as a reference column vector;

select the pair of column vectors in the channel matrix, H, having a highest correlation value and N r /2−2 additional channel vectors having the highest correlation values with respect to the reference column vector as the first sub-matrix; and

select the N r /2 that are not included in the first-matrix as the second sub-matrix.

12 . The RAN node of claim 11 wherein, in order to form the port-sorting matrix, the processing circuitry is further configured to cause the RAN node to:

for each sub-matrix of the two or more sub-matrices:

compute a channel covariance matrix for the sub-matrix;

perform an Eigen Value Decomposition, EVD, on the channel covariance matrix for the sub-matrix;

form the port-sorting matrix based on the eigen vector matrices, U g i , and the eigen value matrices, D g i , obtained via the EVDs performed on the channel covariance matrices for the two or more sub-matrices.

13 . The RAN node of claim 11 wherein the two or more sub-matrices consist of a first sub-matrix and a second sub-matrix, and the port-sorting sub-matrix is computed as:

U

=

[

U

g

1

*

0

0

U

g

2

*

]

[

D

g

1

-

1

2

0

0

D

g

2

-

1

2

]

wherein U g 1 is the eigen vector matrix for the first sub-matrix, D g 1 is the eigen value matrix for the first sub-matrix, U g 2 is the eigen vector matrix for the second sub-matrix, and D g 2 is the eigen value matrix for the second sub-matrix.

14 . The RAN node of claim 9 wherein, in order to apply grouping of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , and port sorting accordingly to obtain the final port-sorted channel matrix, {tilde over (H)} final , the processing circuitry is further configured to cause the RAN node to:

divide the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , into two or more sub-matrices each comprising two or more consecutive column vectors from the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

for each sub-matrix of the two or more sub-matrices of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps :

compute a channel covariance matrix for the sub-matrix of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

perform an EVD on the channel covariance matrix for the sub-matrix of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

form a second port-sorting matrix based on eigen vector matrices, N s k , and the eigen value matrices, L s k , obtained via the EVDs performed on the channel covariance matrices for the two or more sub-matrices of the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps ;

apply the second port-sorting matrix to the re-ordered, port-sorted channel matrix, {tilde over (H)} re-order ps , to provide the final port-sorted channel matrix, {tilde over (H)} final .

15 . The RAN node of claim 9 wherein the processing circuitry is further configured to cause the RAN node to perform layer selection for the particular UE by selecting a first L columns of the final port-sorted channel matrix, {tilde over (H)} final , wherein L is a number of layers for a current rank of the particular UE.

16 . The RAN node of claim 9 wherein the processing circuitry is further configured to cause the RAN node to compute beamforming weights for a downlink transmission to the particular UE based on the final port-sorted channel matrix, {tilde over (H)} final , or the first L columns of the final port-sorted channel matrix, {tilde over (H)} final .

Assignments (1)

Continuity (2)

References Cited (33)