Digital Signal Processing Reference
In-Depth Information
D.3 Discrete-time
Fourier transform
To illustrate the partial fraction expansion of the DTFT, consider the following
rational function:
=
b
m
e
j
m
Ω
+
b
m
−
1
e
j(
m
−
1)
Ω
++
b
1
e
j
Ω
+
b
0
a
n
e
j
n
Ω
+
a
n
−
1
e
j(
n
−
1)
Ω
++
a
1
e
j
Ω
+
a
0
X
(
Ω
)
=
N
(
Ω
)
D
(
Ω
)
,
(D.40)
where the numerator
N
(
Ω
) is a polynomial of degree
m
and the denominator
D
(
Ω
) is a polynomial of degree
n
. An alternative representation for Eq. (D.40)
is obtained by dividing both the numerator and the denominator by e
j
n
Ω
as
follows:
−
j
Ω
++
b
1
e
−
j(
m
−
1)
Ω
+
b
0
e
−
j
m
Ω
X
(
Ω
)
=
N
(
Ω
)
D
(
Ω
)
=
e
j(
m
−
n
)
Ω
b
m
+
b
m
−
1
e
.
a
n
+
a
n
−
1
e
−
j
Ω
++
a
1
e
−
j(
n
−
1)
Ω
+
a
0
e
−
j
n
Ω
X
′
(
ω
)
(D.41)
We need to express Eq. (D.41) in simpler terms using the partial fraction expan-
sion with respect to e
−
j
Ω
. To simplify the factorization process, we substitute
z
=
e
j
Ω
:
X
(
z
)
=
z
(
m
−
n
)
b
m
+
b
m
−
1
z
−
1
++
b
1
z
−
(
m
−
1)
+
b
0
z
−
m
a
n
+
a
n
−
1
z
−
1
++
a
1
z
−
(
n
−
1)
+
a
0
z
−
n
.
(D.42)
The process for the partial fraction expansion of Eq. (D.41) is the same as for
the CTFT and Laplace transform, except that the expansion is performed with
respect to
z
−
1
. Below we illustrate the process with an example.
Example D.8
Using the partial fraction method, calculate the inverse CTFT of the following
function:
2e
j2
Ω
−
5e
j
Ω
e
j2
Ω
−
(4
/
9)e
j
Ω
+
(1
/
27)
.
N
(
Ω
)
D
(
Ω
)
X
(
Ω
)
=
=
(D.43)
Solution
Dividing both the numerator and the denominator of Eq. (D.43) by e
j2
Ω
yields
2
−
5e
−
j
Ω
1
−
(4
/
9)e
−
j
Ω
+
(1
/
27)e
−
2j
Ω
.
X
(
Ω
)
=
Substitute
z
=
e
j
Ω
in the above equation to obtain
2
−
5
z
−
1
1
−
(4
/
9)
z
−
1
+
(1
/
27)
z
−
2
,
with the characteristic equation
X
(
z
)
=
4
9
z
−
1
+
1
27
z
−
2
4
9
z
+
1
27
=
0or
z
2
−
1
−
=
0
,
Search WWH ::
Custom Search