Digital Signal Processing Reference
In-Depth Information
This expression can be rewritten in the following form:
j
P
i
P
j
e
1
2
C
X
ðx
i
m
i
Þðx
j
m
j
Þ
D
i;j
1
j
f
X
ðXÞ¼
p
C
X
i; j ¼
1
;
...
; N;
(4.136)
;
N=
2
ð
2
pÞ
j
j
where
D
i,j
is the cofactor of the matrix
C
X
.
Example 4.6.2
In this example, we demonstrate how the PDF of two jointly
normal random variables (
4.124
) can be derived from the general expression
(
4.136
).
Solution
The covariance matrix
C
X
is
s
1
C
1
;
2
C
X
¼
;
(4.137)
s
2
C
2
;
1
where the covariances are:
C
1
;
2
¼ C
2
;
1
¼ EfX
1
m
1
ÞðX
2
m
2
Þg ¼ r
1
;
2
s
1
s
2
:
(4.138)
The determinant |
C
X
| is given as:
¼ s
1
s
2
C
1
;
2
:
s
1
C
1
;
2
j
C
X
j ¼
(4.139)
s
2
C
2
;
1
Using (
4.138
), the determinant (
4.139
) is expressed as:
j ¼ s
1
s
2
ð
1
r
1
;
2
Þ:
j
C
X
(4.140)
Its cofactors are:
D
1
;
1
¼ s
2
; D
2
;
2
¼ s
1
; D
1
;
2
¼
D
2
;
1
¼C
1
;
2
¼C
2
;
1
¼r
1
;
2
s
1
s
2
:
(4.141)
Placing (
4.138
)-(
4.141
) into (
4.136
), we get:
h
i
1
2
2
ðx
1
m
1
Þ
s
2
s
1
s
2
þ
ðx
2
m
2
Þ
s
1
s
1
s
2
2
r
1
;
2
s
1
s
2
ðx
1
m
1
Þðx
2
m
2
Þ
1
r
1
;
2
1
2
ð
1
r
1
;
2
Þ
f
X
1
;X
2
ðx
1
;x
2
Þ¼
q
exp
s
1
s
2
2
ps
1
s
2
h
i
1
2
s
1
þ
ð
x
2
m
2
Þ
2
ðx
1
m
1
Þ
s
2
2
r
1
;
2
ðx
1
m
1
Þðx
2
m
2
Þ
q
1
r
1
;
2
1
2
ð
1
r
1
;
2
Þ
¼
exp
;
s
1
s
2
2
ps
1
s
2
(4.142)
which is the same expression as in (
4.124
).
Search WWH ::
Custom Search