Digital Signal Processing Reference
In-Depth Information
f
X
ðxÞ
d
x ¼ f
Y
ðyÞ
d
y:
(2.136)
From here, we can write:
f
X
ðxÞ
d
y=
f
Y
ðyÞ ¼
d
x
:
(2.137)
The derivation d
y
/d
x
can be positive as well as negative. However, the density
function can be only positive. To solve this problem, the absolute value of the
derivative d
y
/d
x
must be used in (
2.137
), as demonstrated in the following relation:
f
X
ðxÞ
d
y=
f
Y
ðyÞ ¼
j
:
(2.138)
j
d
x
Note that the left side of (
2.138
) is a function of
y
, while the right side is a
function of
x
. To overcome this problem, we solve (
2.132
) to obtain:
x ¼ g
1
ðyÞ;
(2.139)
in the final expression for the desired density function. This gives us:
x¼g
1
ðyÞ
:
f
X
ðxÞ
d
y=
f
Y
ðyÞ ¼
(2.140)
j
d
x
j
The desired density
f
Y
(
y
) is equal to zero for values of
y
where for
y
there is no
solution (
2.132
).
Example 2.6.1
Show that the linear transformation of the uniform random variable
results in a uniform random variable.
Solution
Let the input variable
X
be in the interval [
x
1
,
x
2
] with the corresponding
density, as shown in Fig.
2.31
.
8
<
1
x
2
x
1
x
1
xx
2
;
for
f
X
ðxÞ ¼
(2.141)
:
0
otherwise
A linear transformation is given as:
y ¼ ax þ b;
(2.142)
where
a
and
b
are constants.
From (
2.142
), we obtain:
x ¼ ðybÞ=a
(2.143)
Search WWH ::
Custom Search