Digital Signal Processing Reference
In-Depth Information
If (
6.152
) is satisfied, we can write
T=
2
ð
1
1
1
T
lim
T!1
xðtÞxðt þ tÞ
d
t ¼
x
1
x
2
f
X
1
X
2
ðx
1
; x
2
;
tÞ
d
x
1
d
x
2
;
(6.153)
1
1
T=
2
where
X
1
and
X
2
are random variables of the process
X
(
t
) in time instants
t
and
t
+
t
,
respectively.
Example 6.7.2
Determine whether the process
XðtÞ¼Y
cos
ðo
0
t þ yÞ
(6.154)
is a mean and autocorrelation ergodic, if
Y
is uniform over [0, 1], and if
o
0
and
y
are
constants.
Solution
The mean value is:
ð
1
EXðtfg¼
y
cos
ðo
0
t þ yÞf
Y
ðyÞ
d
y ¼
cos
ðo
0
t þ yÞ=
2
:
(6.155)
0
The mean value is not a constant and thus the process is not a first-order
stationary and consequently, it is not a mean ergodic.
The autocorrelation function is:
EXðtÞXðt þ tÞ
f
g ¼ EY
cos
ðo
0
t þ yÞY
cos
ðo
0
ðt þ tÞþyÞ
f
g
¼ EfY
2
g
cos
o
0
t þ
cos
ð
2
o
0
t þ o
0
t þ
2
yÞ
½
=
2
1
6
cos
o
0
t þ
cos
ð
2
o
0
t þ o
0
t þ
2
yÞ
¼
½
:
(6.156)
From (
6.156
), the autocorrelation function is a function of time
t
and thus cannot
be equal to a time-averaged autocorrelation function which does not depend on
time
t
. Therefore, the process is not autocorrelation ergodic.
Similarly, for mutually ergodic processes
X
(
t
) and
Y
(
t
), we can make a time and a
statistical cross-correlation function equal as:
t
XðtÞYðt þ tÞ¼XðtÞYðt þ tÞ
¼
Ð
1
1
Ð
1
T=
2
Ð
xðtÞyðt þ tÞ
d
t
:
1
T
x
1
y
2
f
X
1
Y
2
ðx
1
; y
2
;
tÞ
d
x
1
d
y
2
¼
lim
T!1
1
T=
2
(6.157)
Search WWH ::
Custom Search