Digital Signal Processing Reference
In-Depth Information
Given MGF, the moments are derived, by taking derivatives with respect to
s
,
and evaluating the result at
s ¼
0,
s¼
0
;
d
M
X
ð
s
Þ
d
s
m
1
¼
s¼
0
;
d
2
M
X
ð
s
Þ
d
s
2
m
2
¼
s¼
0
;
d
3
M
X
ð
s
Þ
d
s
3
(2.425)
m
3
¼
...
s¼
0
:
d
n
M
X
ð
s
Þ
d
s
n
m
n
¼
When the MGF is evaluated at
s ¼ jo
, the result is the Fourier transform, i.e.,
MGF function becomes the characteristic function. As a difference to MGF, the
characteristic function always converges, and this is the reason why it is often used
instead of MGF, [PEE93, p. 82].
Example 2.8.13
Find the MGF function of the exponential random variable and
find its variance using (
2.425
).
Solution
1
1
e
sx
l
e
lx
d
x ¼ l
1
0
M
X
ðsÞ¼Ef
e
sX
e
sx
f
X
ðxÞ
d
x ¼
e
ðlsÞx
d
x:
g¼
(2.426)
1
0
This integral converges for Re{s}
< l
, where Re{
s
} means the real value of
s
.
From (
2.426
), we easily found:
l
l s
:
M
X
ðsÞ¼
(2.427)
The first and second moments are obtained taking the first and second
derivatives of (
2.427
), as indicated in (
2.428
).
s¼
0
¼
s¼
0
¼
s¼
0
¼
d
M
X
ð
s
Þ
d
s
d
d
s
l
l s
l
ðl sÞ
1
l
;
m
1
¼
2
!
s¼
0
¼
s¼
0
¼
s¼
0
¼
d
2
M
X
ð
s
Þ
d
s
2
d
d
s
l
ðl sÞ
2
l
ðl sÞ
2
l
2
:
m
2
¼
(2.428)
2
3
Search WWH ::
Custom Search