Coefficient calculation method and device for time-domain channel estimation filter
View Patent ↗The present invention provides a method and a device for calculating a coefficient of a time-domain channel estimation filter. The method comprises: acquiring position information and weight information of symbols carrying RS information corresponding to X 1 , X 2 . . . X P ; acquiring position information and weight information of symbols carrying RS information relating to Y 1 , Y 2 , . . . Y Q ; calculating an autocorrelation matrix of a vector I in accordance with the position information and the weight information, wherein I=[X 1 , X 2 , . . . X P , Y 1 , Y 2 , . . . Y Q ]; calculating a crosscorrelation matrix of a vector O and the vector I in accordance with the position information and the weight information, wherein O=[Y 0 ], and Y 0 is a time-domain channel ideal value of the symbol on which the channel estimation is currently performed; and calculating an inverse matrix product of the crosscorrelation matrix and the autocorrelation matrix, thereby obtaining a coefficient vector corresponding to the symbol on which the channel estimation is currently performed. According to the present invention, it is able to calculate an optimal filter coefficient.
1. A method for calculating a coefficient of a time-domain channel estimation filter for use in a Long Term Evolution (LTE) system, comprising:
acquiring position information and weight information of symbols carrying reference signal (RS) information corresponding to X 1 , X 2 . . . X P , wherein X 1 , X 2 . . . X P are P frequency-domain channel estimation values of the symbols closest to a currently-input symbol, and P is an integer greater than 0;
acquiring position information and weight information of symbols carrying RS information relating to Y 1 , Y 2 , . . . Y Q , wherein Y 1 , Y 2 , . . . Y Q are Q time-domain channel estimation values of the symbols closest to a symbol on which the channel estimation is currently performed, and Q is an integer greater than 0;
calculating an autocorrelation matrix of a vector I in accordance with the position information and the weight information, wherein I=[X 1 , X 2 . . . X P , Y 1 , Y 2 , . . . Y Q ];
calculating a crosscorrelation matrix of a vector O and the vector I in accordance with the position information and the weight information, wherein O=[Y 0 ], and Y 0 is a time-domain channel ideal value of the symbol on which the channel estimation is currently performed; and
calculating an inverse matrix product of the crosscorrelation matrix and the autocorrelation matrix, thereby obtaining a coefficient vector corresponding to the symbol on which the channel estimation is currently performed.
2. The method according to claim 1 , wherein the acquiring position information and weight information of symbols carrying RS information corresponding to X 1 , X 2 . . . X P comprises:
setting position information D (X i ) and weight information W (X i ) corresponding to X i to 0 during initialization, wherein i=1, . . . P; and
when a new symbol is input, judging whether or not the symbol carries the RS information, if yes, updating D (X i ) to D (X i-1 ) and updating W (X i ) to W (X i-1 ) from i=P, and when i=1, updating D (X 1 ) to n and updating W (X 1 ) to 1, n being a serial number of the symbol, otherwise, not updating the position information and the weight information.
3. The method according to claim 1 , wherein the acquiring position information and weight information of symbols carrying RS information corresponding to X 1 , X 2 . . . X P comprises:
setting position information D (X i ) and weight information W (X i ) corresponding to X i to 0 during initialization, wherein i=1, . . . P; and
when a new symbol is input, updating D (X i ) to D (X i-1 ) and updating W (X i ) to W (X i-1 ) from i=P, judging whether or not the symbol carries the RS information when i=1, if yes, updating D (X 1 ) to n and updating W (X 1 ) to 1, n being a serial number of the symbol, otherwise, not updating D (X 1 ) and W (X 1 ) .
4. The method according to claim 1 , wherein an autocorrelation matrix R II of the vector I is calculated by the following equations:
R
II
=
E
[
II
H
]
=
E
[
X
1
X
1
*
X
1
X
2
*
…
X
1
Y
Q
*
X
2
X
1
*
X
2
X
2
*
…
X
2
Y
Q
*
⋮
⋮
⋱
⋮
X
Q
X
1
*
X
Q
X
2
*
…
Y
Q
Y
Q
*
]
E
[
X
k
X
m
*
]
=
(
E
[
X
k
*
X
m
]
)
*
=
W
(
X
k
)
(
W
(
X
m
)
)
*
R
(
τ
D
(
X
k
)
,
D
(
X
m
)
)
,
k
,
m
=
1
,
…
P
E
[
Y
k
Y
m
*
]
=
(
E
[
Y
k
*
Y
m
]
)
*
=
∑
i
=
1
length
(
D
(
Y
k
)
)
∑
j
=
1
length
(
D
(
Y
m
)
)
W
(
Y
k
)
(
i
)
(
W
(
Y
m
)
(
j
)
)
*
R
(
τ
D
(
Y
k
)
(
i
)
,
D
(
Y
m
)
(
j
)
)
,
k
,
m
=
1
,
…
Q
E
[
Y
k
X
m
*
]
=
(
E
[
Y
k
*
X
m
]
)
*
=
∑
i
=
1
length
(
D
(
Y
k
)
)
W
(
Y
k
)
(
i
)
(
W
(
X
m
)
)
*
R
(
τ
D
(
Y
k
)
(
i
)
,
D
(
X
m
)
)
,
k
=
1
,
…
Q
,
m
=
1
,
…
P
wherein E[•] represents mathematical expectation operation, D (X m ) is the position information corresponding to X m , W (X m ) is the weight information corresponding to X m , m=1, . . . P, D (Y k ) is the position information corresponding to Y k , W (Y k ) is the weight information corresponding to Y k , k=1, . . . Q, W (Y k ) (j) is the j th element of W (Y k ) , D (Y k ) (j) is the j th element of D (Y k ) , length(•) is the number of elements when the vector is calculated, R(τ i,j ) is a correlation function determined in accordance with channel information and system information, and τ i,j is the time distance between the i th symbol and the j th symbol.
5. The method according to claim 1 , wherein a crosscorrelation matrix R OI between the vector O and the vector I is calculated by the following equations:
R
OI
=
E
[
OI
H
]
=
E
[
Y
0
X
1
*
Y
0
X
2
*
⋮
Y
0
Y
Q
*
]
E
[
Y
0
Y
p
*
]
=
(
W
(
X
p
)
)
*
R
(
τ
n
-
L
,
D
(
X
p
)
)
,
p
=
1
,
…
P
E
[
Y
0
Y
q
*
]
=
∑
i
=
1
length
(
D
(
Y
q
)
)
(
W
(
Y
q
)
(
i
)
)
*
R
(
τ
n
-
L
,
D
(
Y
q
)
(
i
)
)
,
q
=
1
,
…
Q
wherein E[•] represents mathematical expectation operation, D (X p ) is the position information corresponding to X p , W (X p ) is the weight information corresponding to X p , p=1, . . . P, D (Y q ) is the position information corresponding to Y q , W (Y q ) is the weight information corresponding to Y q , q=1, . . . Q, W (Y q ) (i) is the i th element of W (Y q ) , D (Y q ) (i) is the i th element of D (Y q ) , length(•) is the number of elements when the vector is calculated, R(τ i,j ) is a correlation function determined in accordance with channel information and system information, τ i,j the time distance between the i th symbol and the j th symbol, n−L is the serial number of the symbol on which the channel estimation is currently performed, n is the serial number of the currently-input symbol, and L is a time delay, i.e., the number of symbols between input and output of the filter.
6. The method according to claim 4 , wherein the correlation function is:
R (τ i,j )= J 0 (2 πf max τ i,j )
wherein J 0 (•) represents a first-class 0-order Bessel function, f max represents a maximum Doppler shift, τ i,j =T sym |i−j|, and T sym represents a time length of a symbol.
7. The method according to claim 1 , wherein when the autocorrelation matrix is not of full rank, the method further comprises:
updating the autocorrelation matrix to a full-rank matrix by means of dimension reduction, and updating the crosscorrelation matrix correspondingly,
and wherein the calculating an inverse matrix product of the crosscorrelation matrix and the autocorrelation matrix comprises calculating an inverse matrix product of the updated crosscorrelation matrix and the updated autocorrelation matrix.
8. The method according to claim 7 , further comprising:
normalizing the coefficient vector.
9. The method according to claim 1 , wherein the acquiring the position information and the weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q comprises:
acquiring the position information and weight information of the symbols carrying the RS information relating to Y o Y 2 , . . . Y Q corresponding to the symbol on which the channel estimation is currently performed, wherein the position information and the weight information are obtained after the coefficient calculation on a previous symbol, and during initialization, the position information and the weight information are blank, and
after obtaining the coefficient vector corresponding to the symbol on which the channel estimation is currently performed, the method further comprises updating the position information and the weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q for the coefficient calculation on a next symbol.
10. The method according to claim 9 , wherein the updating the position information and the weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q comprises:
making W (temp) =a 1 W (Y 1 ) and D (temp) =D (Y 1 ) ;
starting from q=2, updating W (temp) as follows until q=Q:
W (temp) (1: length(W (Y q ) ))=W (temp) (1: length(W (Y q ) ))+a q W (Y q )
starting from p=1, updating W (temp) and D (temp) as follows until p=P:
if D (X p ) belongs to D (temp) , making W (temp) (i)=W (temp) (i)+b p , i being the serial number of the element in D (temp) that is equal to D (X p ) , and
if D (X p ) does not belong to D (temp) , making W (temp) =[W (temp) b p ] and D (temp) =[D (temp) D (X p ) ]; and
starting from q=Q, updating W (Y q ) to W (Y q-1 ) , updating D (Y q ) to D (Y q-1 ) , and when q=1, updating W (Y 1 ) to W (temp) and updating D (Y 1 ) to D (temp) ,
wherein D (X p ) represents the position information corresponding to X p , p=1, . . . P, D (Y q ) represents the position information corresponding to Y q , W (Y q ) represents the weight information corresponding to Y q , q=1, . . . Q, [b 1 , . . . b P , a 1 , . . . a Q ] represents the coefficient vector corresponding the symbol on which the channel estimation is currently performed, and length(•) represents number of elements for calculating the vector.
11. The method according to claim 1 , further comprising:
calculating a distance vector between a symbol corresponding to a valid element in I and the symbol on which the channel estimation is currently performed;
determining a scenario to which the symbol on which the channel estimation is currently performed belongs in accordance with the distance vector; and
after the coefficient calculation enters a stable state, recording the calculated coefficient vector as a filter coefficient for the corresponding scenario.
12. A device for calculating a coefficient of a time-domain channel estimation filter for use in a Long Term Evolution (LTE) system, comprising:
a first acquiring module, configured to acquire position information and weight information of symbols carrying reference signal (RS) information corresponding to X 1 , X 2 . . . X P , wherein X 1 , X 2 . . . X P are P frequency-domain channel estimation values of the symbols closest to a currently-input symbol, and P is an integer greater than 0;
a second acquiring module, configured to acquire position information and weight information of symbols carrying RS information relating to Y 1 , Y 2 , . . . Y Q , wherein Y 1 , Y 2 , . . . Y Q are Q time-domain channel estimation values of the symbols closest to a symbol on which the channel estimation is currently performed, and Q is an integer greater than 0;
an autocorrelation calculating module, configured to calculate an autocorrelation matrix of a vector I in accordance with the position information and the weight information, wherein I=[X 1 , X 2 . . . X P , Y 1 , Y 2 , . . . Y Q ];
a crosscorrelation calculating module, configured to calculate a crosscorrelation matrix of a vector O and the vector I in accordance with the position information and the weight information, wherein O=[Y 0 ], and Y 0 is a time-domain channel ideal value of the symbol on which the channel estimation is currently performed; and
a filter coefficient calculating module, configured to calculate an inverse matrix product of the crosscorrelation matrix and the autocorrelation matrix, thereby to obtain a coefficient vector corresponding to the symbol on which the channel estimation is currently performed.
13. The device according to claim 12 , wherein the first acquiring module is further configured to:
set position information D (X i ) and weight information W (X i ) corresponding to X i to 0 during initialization, wherein i=1, . . . P; and
when a new symbol is input, judge whether or not the symbol carries the RS information, if yes, update D (X i ) to D (X i-1 ) and update W (X i ) to W (X i-1 ) from i=P, and when i=1, update D (X 1 ) to n and update W (X 1 ) to 1, n being a serial number of the symbol, otherwise, not update the position information and the weight information.
14. The device according to claim 12 , wherein the first acquiring module is further configured to:
set position information D (X i ) and weight information W (X i ) corresponding to X i to 0 during initialization, wherein i=1, . . . P; and
when a new symbol is input, update D (X i ) to D (X i-1 ) and update W (X i ) to W (X i-1 ) from i=P, judge whether or not the symbol carries the RS information when i=1, if yes, update D (X 1 ) to n and update W (X 1 ) to 1, n being a serial number of the symbol, otherwise, not update D (X 1 ) and W (X 1 ) .
15. The device according to claim 12 , wherein the autocorrelation calculating module is further configured to calculate an autocorrelation matrix R II of the vector I by the following equations:
R
II
=
E
[
II
H
]
=
E
[
X
1
X
1
*
X
1
X
2
*
…
X
1
Y
Q
*
X
2
X
1
*
X
2
X
2
*
…
X
2
Y
Q
*
⋮
⋮
⋱
⋮
Y
Q
X
1
*
Y
Q
X
2
*
…
Y
Q
Y
Q
*
]
E
[
X
k
X
m
*
]
=
(
E
[
X
k
*
X
m
]
)
*
=
W
(
X
k
)
(
W
(
X
m
)
)
*
R
(
τ
D
(
X
k
)
,
D
(
X
m
)
)
,
k
,
m
=
1
,
…
P
E
[
Y
k
Y
m
*
]
=
(
E
[
Y
k
*
Y
m
]
)
*
=
∑
i
=
1
length
(
D
(
Y
k
)
)
∑
j
=
1
length
(
D
(
Y
m
)
)
W
(
Y
k
)
(
i
)
(
W
(
Y
m
)
(
j
)
)
*
R
(
τ
D
(
Y
k
)
(
i
)
,
D
(
Y
m
)
(
j
)
)
,
k
,
m
=
1
,
…
Q
E
[
Y
k
X
m
*
]
=
(
E
[
Y
k
*
X
m
]
)
*
=
∑
i
=
1
length
(
D
(
Y
k
)
)
W
(
Y
k
)
(
i
)
(
W
(
X
m
)
)
*
R
(
τ
D
(
Y
k
)
(
i
)
,
D
(
X
m
)
)
,
k
=
1
,
…
Q
,
m
=
1
,
…
P
wherein E[•] represents mathematical expectation operation, D (X m ) is the position information corresponding to X m , W (X m ) is the weight information corresponding to X m , m=1, . . . P, D (Y k ) is the position information corresponding to Y k , W (Y k ) is the weight information corresponding to Y k , k=1, . . . Q, W (Y k ) (j) is the j th element of W (Y k ) , D (Y k ) (j) is the j th element of D (Y k ) , length(•) is the number of elements when the vector is calculated, R(τ i,j ) is a correlation function determined in accordance with channel information and system information, and τ i,j is the time distance between the i th symbol and the j th symbol.
16. The device according to claim 12 , wherein the crosscorrelation calculating module is further configured to calculate a crosscorrelation matrix R OI between the vector O and the vector I by the following equations:
R
OI
=
E
[
OI
H
]
=
E
[
Y
0
X
1
*
Y
0
X
2
*
⋮
Y
0
Y
Q
*
]
E
[
Y
0
X
p
*
]
=
(
W
(
X
p
)
)
*
R
(
τ
n
-
L
,
D
(
X
p
)
)
,
p
=
1
,
…
P
E
[
Y
0
Y
q
*
]
=
∑
i
=
1
length
(
D
(
Y
q
)
)
(
W
(
Y
q
)
(
i
)
)
*
R
(
τ
n
-
L
,
D
(
Y
q
)
(
i
)
)
,
q
=
1
,
…
Q
and wherein E[•] represents mathematical expectation operation, D (X p ) is the position information corresponding to X 9 , W (X p ) is the weight information corresponding to X P , p=1, . . . P, D (Y q ) is the position information corresponding to Y q , W (Y q ) is the weight information corresponding to Y q , q=1, . . . Q, W (Y q ) (i) is the i th element of W (Y q ) , D (Y q ) (i) is the i th element of D (Y q ) , length(•) is the number of elements when the vector is calculated, R(τ i,j ) is a correlation function determined in accordance with channel information and system information, τ i,j is the time distance between the i th symbol and the j th symbol, n−L is the serial number of the symbol on which the channel estimation is currently performed, n is the serial number of the currently-input symbol, and L is a time delay, i.e., the number of symbols between input and output of the filter.
17. The device according to claim 15 , wherein the correlation function is:
R (τ i,j )= J 0 (2 πf max τ i,j )
wherein J 0 (•) represents a first-class 0-order Bessel function, f max represents a maximum Doppler shift, τ i,j =T sym |i−j|, and T sym represents a time length of a symbol.
18. The device according to claim 12 , wherein the second acquiring module is further configured to:
acquire the position information and weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q corresponding to the symbol on which the channel estimation is currently performed, wherein the position information and the weight information are obtained after the coefficient calculation on a previous symbol, and during initialization, the position information and the weight information are blank; and
after obtaining the coefficient vector corresponding to the symbol on which the channel estimation is currently performed, update the position information and the weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q for the coefficient calculation on a next symbol.
19. The device according to claim 18 , wherein the second acquiring module is further configured to update the position information and the weight information of the symbols carrying the RS information relating to Y 1 , Y 2 , . . . Y Q by:
making W (temp) =a 1 W (Y 1 ) and D (temp) =D (Y 1 ) ;
starting from q=2, updating W (temp) as follows until q=Q:
W (temp) (1: length(W (Y q ) ))=W (temp) (1: length(W (Y q ) ))+a q W (Y q )
starting from p=1, updating W (temp) and D (temp) as follows until p=P:
if D (X p ) belongs to D (temp) , making W (temp) (i)=W (temp) (i)+b p , i being the serial number of the element in D (temp) that is equal to D (X p ) , and
if D (X p ) does not belong to D (temp) , making W (temp) =[W (temp) b p ] and D (temp) =[D (temp) D (X p ) ]; and
starting from q=Q, updating W (Y q ) to W (Y q-1 ) , updating D (Y q ) to D (Y q-1 ) , and when q=1, updating W (Y 1 ) to W (temp) and updating D (Y 1 ) to D (temp) ,
and wherein D (Y p ) represents the position information corresponding to X P , p=1, . . . P, D (Y q ) represents the position information corresponding to Y q , W (Y q ) represents the weight information corresponding to Y q , q=1, . . . Q, [b 1 , . . . b P , a 1 , . . . a Q ] represents the coefficient vector corresponding the symbol on which the channel estimation is currently performed, and length(•) represents number of elements for calculating the vector.
20. The device according to claim 12 , further comprises a filter coefficient selecting module configured to:
calculate a distance vector between a symbol corresponding to a valid element in I and the symbol on which the channel estimation is currently performed;
determine a scenario to which the symbol on which the channel estimation is currently performed belongs in accordance with the distance vector; and
after the coefficient calculation enters a stable state, record the calculated coefficient vector as a filter coefficient for the corresponding scenario.