Digital Signal Processing Reference
In-Depth Information
The normal variable
Y
has the following PDF:
2
e
ð
y
m
Y
Þ
1
2
p
2
s
Y
f
Y
ðyÞ¼
p
:
(5.6)
s
Y
Placing (
5.6
) and (
5.5
) into (
5.4
), we obtain the density function of the lognormal
variable:
2
e
ð
ln
x
m
Y
Þ
2
s
Y
p
2
p
1
for
x
0
;
f
X
ðxÞ¼
(5.7)
s
Y
x
f
X
ðxÞ¼
0
for
x <
0
Note that
m
Y
and
s
Y
respectively, are the mean value and variance of the normal
random variable
Y
, and not of the lognormal variable
X
. Figure
5.1
shows the
lognormal variables (a) and densities (b), for different values of
s
Y
and for
m
Y
¼
0.01. For small values of
s
Y
,(
s
Y
0.3), the lognormal PDF becomes more
symmetric, while for
s
Y
>
0.3, the PDF becomes asymmetric.
5.1.2 Distribution Function
Using the definitions (
2.61
) and (
5.7
), the distribution function of a lognormal
variable is:
2
e
ð
ln
x
m
Y
Þ
ð
x
ð
x
1
2
ps
Y
2
s
Y
F
X
ðxÞ¼PfX xg¼
f
X
ðxÞdx ¼
p
d
x:
(5.8)
x
1
1
However, we can express the distribution (
5.8
) in terms of the normal distribu-
tion and thus make possible the use of erf function to more easily derive the
lognormal distribution, as shown in the following equation:
2
e
ð
y
m
Y
Þ
ð
ð
ln
x
ln
x
1
2
ps
Y
2
s
Y
F
X
ðxÞ¼PfX xg¼
f
Y
ðyÞdy ¼
p
d
y:
(5.9)
1
1
Using the expressions for erf function (
4.29
) we arrive at:
8
<
ln
x
m
Y
1
2
1
þ
erf
p
s
Y
x
0
;
for
F
X
ðxÞ¼
(5.10)
:
0
otherwise
:
As an example, Fig.
5.2
shows the distribution function for
m
Y
¼
0.01 and
s
Y
¼
1.
Search WWH ::
Custom Search