B.5 MultiAntenna channel models
36.1413GPPBase Station (BS) conformance testingEvolved Universal Terrestrial Radio Access (EUTRA)Release 17TS
The MIMO channel correlation matrices defined in B.5 apply for the antenna configuration using uniform linear arrays at both UE and eNodeB.
B.5.1 Definition of MIMO Correlation Matrices
Table B.5.11 defines the correlation matrix for the eNodeB:
Table B.5.11: eNodeB correlation matrix
One antenna 
Two antennas 
Four antennas 

eNode B Correlation 
Table B.5.12 defines the correlation matrix for the UE:
Table B.5.12: UE correlation matrix
One antenna 
Two antennas 
Four antennas 

UE Correlation 
Table B.5.13 defines the channel spatial correlation matrix. The parameters α and β in Table B.5.13 defines the spatial correlation between the antennas at the eNodeB and UE.
Table B.5.13: correlation matrices
1×2 case 

2×2 case 

2×4 case 

4×4 case 
For cases with more antennas at either eNodeB or UE or both, the channel spatial correlation matrix can still be expressed as the Kronecker product of and according to .
B.5.2 MIMO Correlation Matrices at High, Medium and Low Level
The and for different correlation types are given in Table B.5.21.
Table B.5.21: Correlation for High Medium and Low Level
Low correlation 
Medium Correlation 
High Correlation 

0 
0 
0.9 
0.3 
0.9 
0.9 
The correlation matrices for high, medium and low correlation are defined in Table B.5.22, B.5.23 and B.5.24 as below.
The values in Table B.5.22 have been adjusted for the 2×4 and 4×4 high correlation cases to insure the correlation matrix is positive semidefinite after roundoff to 4 digit precision. This is done using the equation:
Where the value “a” is a scaling factor such that the smallest value is used to obtain a positive semidefinite result. For the 2×4 high correlation case, a=0.00010. For the 4×4 high correlation case, a=0.00012.
The same method is used to adjust the 4×4 medium correlation matrix in Table B.5.23 to insure the correlation matrix is positive semidefinite after roundoff to 4 digit precision with a =0.00012.
Table B.5.22: MIMO correlation matrices for high correlation
1×2 case 

2×2 case 

2×4 case 

4×4 case 
Table B.5.23: MIMO correlation matrices for medium correlation
1×2 case 
N/A 
2×2 case 

2×4 case 

4×4 case 
Table B.5.24: MIMO correlation matrices for low correlation
1×2 case 

1×4 case 

2×2 case 

2×4 case 

4×4 case 
In Table B.5.24, is a identity matrix.
NOTE: For completeness, the 1×2 cases were defined for high, medium and low correlation but for Rel8 onwards for 1Tx, performance requirements exist only for low correlation.