You are on page 1of 23

Dr. Mahmoud M.

Al-Husari

Signals and Systems


Discrete Time

This set of lecture notes are never to be considered as a substitute


to the textbook recommended by the lecturer.
ii
Contents

7 Discrete Time Signals and Systems 1


7.1 Exponential Sequences . . . . . . . . . . . . . . . . . . . . . . . . 1
7.2 Discrete-Time Sinusoids . . . . . . . . . . . . . . . . . . . . . . . 3
7.3 Fourier Analysis of Discrete Time Systems . . . . . . . . . . . . . 4
7.3.1 The Discrete Time Fourier Series (DTFS) . . . . . . . . . 4
7.3.2 The Discrete Time Fourier Transform . . . . . . . . . . . 5
7.3.3 Fourier Transform of DT Periodic Sequences . . . . . . . 9
7.3.4 Properties of the DTFT . . . . . . . . . . . . . . . . . . . 9
7.4 Relationship between CTFT and DTFT . . . . . . . . . . . . . . 10
7.5 The Discrete Fourier Transform . . . . . . . . . . . . . . . . . . . 13
7.6 Examples of the Use of The DFT . . . . . . . . . . . . . . . . . . 16

iii
iv CONTENTS
Chapter 7
Discrete Time Signals and
Systems

In the preceding chapters, we discussed mainly techniques for the analysis of


continuous time signals and systems. In Chapter 3 we did some system analysis
for discrete time systems, concepts such as impulse response and discrete time
convolution was introduced. In Chapter 2 we introduced elementary discrete
time signals such as the delta function, unit step function, energy and power
signals, we now discuss exponentials, sine and cosine waveforms.

7.1 Exponential Sequences


A discrete time exponential signal is given by

x[n] = Cαn (7.1)

where, in general, C and α are complex numbers. The exponential term αn can
also be expressed as

αn = eβn (α = eβ or β = ln α) (7.2)

For example, 5n = e1.609n because 5 = e1.609 . Similarly, e3n = (e3 )n =


(20.086)n . We now consider different cases of the discerete exponential sig-
nal in (7.1).
For C and α real, x[n] increases with increasing n if |α| > 1. Similarly, if |α| < 1,
we have a decreasing exponential. Suppose β in (7.2) is purely imaginary, so
that we have
x[n] = CejΩn
The exponential ejΩn is plotted in the complex plane for various values of n, as
illustrated in Figure 7.1 The function x[n] = ejΩn takes on values ej0 , ejΩ , ej2Ω ,
ej3Ω , · · · at n = 0, 1, 2, 3, · · · , respectively. Note that

ejΩn = rejθ , r = 1, and θ = nΩ

1
2 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

Locus of ejΩn
n=3
n=2
n=1
n=0

Figure 7.1: Locus of ejΩn .

This fact shows that the magnitude and angle of ejΩn are 1 and nΩ, respectively.
Therefore, the points ej0 , ejΩ , ej2Ω , ej3Ω , · · · , ejΩn , · · · lie on a circle of unit
radius at angles 0, Ω, 2Ω, 3Ω, · · · , nΩ, · · · respectively. For each unit increase
in n, the function x[n] = ejΩn moves along the unit circle counterclockwise by
an angle Ω. Therefore, the locus of ejΩn may be viewed as a phasor rotating
counterclockwise at a uniform speed of Ω radians per unit sample interval.
This signal is closely related to the discrete time sinusoids using Euler’s formula,
ejΩn = cos Ωn + j sin Ωn
If we assume n is dimensionless, then the frequency of ejΩn is Ω and has the
units of radians. We recall that the continuous time signal x(t) = ejωt is periodic
with period T = 2π/ω for any ω. The interseting question here: is a discrete
time exponential, ejΩn , always periodic for any Ω. A discrete time signal x[n]
is said to be periodic if
x[n] = x[n + N ]
Therefore, for the case x[n] = ejΩn we must have
ejΩn = ejΩ(n+N ) = ejΩn ejΩN
This result is possible only if
ejΩN = 1
For this to hold, ΩN must be an integer multiple of 2π, so that
ΩN = 2πm m = 0, ±1, ±2, · · ·
or
Ω m
=
2π N
Because both m and N are integers, thus, x[n] is periodic only if Ω/2π is a
rational number. To compute the fundamental period N0 , we must choose
the smallest value of m that will make m(2π/Ω) an integer. For example, if
Ω = 4π/17, then the smallest value of m that will make
2π 17
m =m
Ω 2
7.2. DISCRETE-TIME SINUSOIDS 3

an integer is 2. Therefore
2π 17
N0 = m =2 = 17
Ω 2
Let Example 7.1
x[n] = ej ( )n

9

Find the fundamental period.

 Solution First we have


Ω m 7
= =
2π N 18
Thus, the sequence is periodic and the fundamental period, obtained by choosing
m = 7, is given by N = 18. 

Let Example 7.2


x[n] = ej ( 9 )n
7

We have
Ω 7
=
2π 18π
which is not rational. Thus the sequence is not periodic. 

7.2 Discrete-Time Sinusoids


A continuous-time sinusoid cos ωt has a unique waveform for every value of ω in
the range 0 tp ∞. Increasing ω results in a sinusoid of ever increasing frequency.
Such is not the case for the discrete-time sinusoid cos Ωn because

cos(Ω ± 2πk)n = cos(Ωn ± 2πkn)

TO BE CONTINUED

Figure 7.2: Highest oscillation rate in a discrete-time sinusoid occurs at Ω = π.


4 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

7.3 Fourier Analysis of Discrete Time Systems


In chapters 4, 5, and 6, we studied the ways to representing a continuous time
signal as a sum of sinusoids or exponentials. In this section we shall discuss
similar development for discrete time signals. We first represent a periodic x[n]
as a Fourier series formed by a discrete time exponential and its harmonics.

7.3.1 The Discrete Time Fourier Series (DTFS)


A continuous time periodic signal of period T can be represented using the
exponential form of the Fourier series consisting of exponentials ej0t , e±jω0 t ,
e±j2ω0 t , e±j3ω0 t , · · · where ω0 = 2π/T is the fundamental frequency. In analogy
with our representation of periodic signals in continuous time, we can expect
that we can obtain such a representation in terms of the exponentials ej0n ,
e±jΩ0 n , e±j2Ω0 n , · · · , e±jkΩ0 n , · · · , and so on. That is we seek a representation
for x[n], of the form
X∞
x[n] = ak ejkΩ0 n (7.3)
k=−∞
It would appear that we need infinite number of harmonics, however it turns
out that we have to include only N terms in the summation on the right side in
(7.3). The reason being that discrete time exponentials of frequencies seperated
by integer multiples of 2π are identical as discussed next.

Let xk [n] define the set of functions


xk [n] = ejkΩ0 n k = 0, ±1, ±2, · · ·
with Ω0 = 2π/N . Therefore, the set is
{· · · , e−jΩ0 n , 1, ejΩ0 n , ej2Ω0 n , · · · , ejkΩ0 n , · · · , ej(k+N )Ω0 n , · · · }
where ejkΩ0 n is the k th harmonic to the fundamental signal ejΩ0 n . Consider the
(k + N )th harmonic
xk+N [n] = ej(k+N )Ω0 n = ejkΩ0 n ejN Ω0 n = ejkΩ0 n ej2πn = xk [n]
Here we have used the fact that ej2πn = 1. Hence there are only N distinct
waveforms in the set given above. These correspond to the frequencies Ωk = kΩ0
for k = 0, 1, · · · , N − 1. The first harmonic is identical to the N + 1 harmonic,
the second hamrmonic is identical to the N + 2 harmonic, and so on. The
consequence of this result on the Fourier series representation of a discrete time
periodic signal x[n] in (7.3) is that the summation is only needed over any
N consecutive values of k, for example k = 0, 1, 2, · · · , N − 1. Therefore, the
representation for x[n] can be written as
X
x[n] = ak ejkΩ0 n (7.4)
k=<N >

where k =< N > indicates the range of summation of any N succesive integers.

The coefficients ak in the Fourier series (7.4) can be computed using (given here
without proof)
1 X
ak = x[n]e−jkΩ0 n (7.5)
N
n=<N >
7.3. FOURIER ANALYSIS OF DISCRETE TIME SYSTEMS 5

Since xk [n] = xk+N [n], it is clear that


ak+N = ak
and if x[n] is a real periodic signal then
ak = a∗−k+N
Let x[n] be a periodic extenstion of the sequence Example 7.3
{2, −1, 1, 2}
Determine the Fourier coefficients ak .

 Solution The period is N = 4, so that Ω0 = π/2, from (7.5) we have


3
1X
ak = x[n]e−jk(π/2)n
4 n=0
Hence,
3
1X 1
a0 = x[n] = (2 − 1 + 1 + 2) = 1
4 n=0 4
3 3
1X 1X
a1 = x[n]e−j(π/2)n = x[n](−j)n
4 n=0 4 n=0
3 3
1X 1X
a2 = x[n]e−j2(π/2)n = x[n](−1)n
4 n=0 4 n=0
3
1X
a3 = x[n]e−j3(π/2)n
4 n=0
TO BE CONTINUED 

7.3.2 The Discrete Time Fourier Transform


We now consider the frequency domain representation of discrete-time signals
that are not necessarily periodic. For continuous-time signals, we obtained such
a representation by defining the Fourier transform of a signal x(t) as
Z ∞
X(ω) = F[x(t)] = x(t)e−jωt dt
−∞

For discrete-time signals, we consider an analogous definition of the Fourier


transform as
X∞
X(Ω) = F[x[n]] = x[n]e−jΩn (7.6)
n=−∞
It is important to note here that X(Ω) is periodic with period 2π. According
to (7.6) it follows that

X
X(Ω + 2π) = x[n]e−j(Ω+2π)n
n=−∞
X∞
= e−jΩn e−j2πn = X(Ω)
n=−∞
6 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

As a consequence, we have to consider values of Ω only over the range [0, 2π].
Furthermore, Ω is continuous, this fact makes the spectrum of X(Ω) continuous
and periodic function of Ω with period 2π.

To find the inverse relation between X(Ω) and x[n], we replace the variable n
in (7.6) by m to get
X∞
X(Ω) = x[m]e−jΩm
m=−∞
jΩn
Now multiply both sides by e and integrate over the range [0, 2π] to get
Z Z X∞
X(Ω)ejΩn dΩ = x[m]ejΩ(n−m) dΩ
2π 2π m=−∞

Interchanging the orders of summation and integration then gives


Z ∞
X Z
X(Ω)ejΩn dΩ = x[m] ejΩ(n−m) dΩ
2π m=−∞ 2π

It can be verified that


Z (
2π, n=m
ejΩ(n−m) dΩ =
2π 0, n 6= m

We can, therefore, write


Z
1
x[n] = X(Ω)ejΩn dΩ
2π 2π

We consider a few examples.

Example 7.4 Find the DTFT of


x[n] = αn u[n], |α| < 1
 Solution For this example

X
X(Ω) = αn e−jΩn
n=0

X n
= αe−jΩ
n=0

This is a geometric progression with a common ration αe−jΩ . Therefore,


1 1
X(Ω) = =
1 − αe−jΩ 1 − α cos Ω + jα sin Ω
provided that |αe−jΩ | < 1 which is the case since |α| < 1. The magnitude is
given by
1
|X(Ω)| = q
2
(1 − α cos Ω)2 + (α sin Ω)
1
=√
1 + α2 − 2α cos Ω
7.3. FOURIER ANALYSIS OF DISCRETE TIME SYSTEMS 7

and the phase by


 
α sin Ω
∠X(Ω) = − tan−1
1 − α cos Ω

Figure 7.3 shows x[n] = αn u[n] and its spectra for α = 0.8. Observe that the

x[n]

|X(Ω)|

∠X(Ω)

Figure 7.3: Fourier spectra of Emaple 5.4.

frequency spectra are continuous and periodic functions of Ω with the period
2π. The magnitude spectrum is an even function and the phase specturm is an
odd function of Ω. 

Consider the function Example 7.5


x[n] = δ[n]

 Solution Its Fourier transform is



X
X(Ω) = δ[n]e−jΩn = 1
n=−∞

and the Fourier transform of δ[n − k] is e−jkΩ . 

Let Example 7.6


|n|
x[n] = α , |α| < 1
8 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

 Solution We obtain its Fourier transform as


X
X(Ω) = α|n| e−jΩn
n=−∞
−1
X ∞
X
= α−n e−jΩn + αn e−jΩn
n=−∞ n=0
∞ ∞
X n X n
= −1 + αejΩ + αe−jΩ
n=0 n=0
1 − α2
=
1 − 2α cos Ω + α2

In this case, X(Ω) is real, so that the phase is zero. 

Example 7.7 Find the inverse Fourier transform of the rectangular pulse spectrum shown in
Figure 7.4.

|X(Ω)|

x[n]

Figure 7.4: Inverse DTFT of a periodic rect spectrum 5.7.

 Solution Using
Z
1
x[n] = X(Ω)ejΩn dΩ
2π 2π
Z π/4
1
= ejΩn dΩ
2π −π/4
π/4
1 jΩn
= e
j2πn −π/4

sin π4 n

1 n
= = sinc 
πn 4 4
7.3. FOURIER ANALYSIS OF DISCRETE TIME SYSTEMS 9

7.3.3 Fourier Transform of DT Periodic Sequences


For continuous-time periodic signals, we obtained the Fourier transform of a
periodic signal x(t) as

X
X(ω) = 2πcn δ(ω − nω0 )
n=−∞

For discrete-time periodic signals, we consider an analogus definition of the


Fourier transform as

X
X(Ω) = 2πak δ(Ω − kΩ0 )
k=−∞

Find the discrete-time Fourier transform for the periodic signal Example 7.8
 
3π π π 
x[n] = 2 cos n+ + 4 sin n
8 3 2
 Solution First we need to determine the DTFS. The fundamental frequency
is Ω0 = π/8. Therefore, only the third and fourth harmonics exist in the DTFS
spectrum. Using Euler identities we deduce



 −2/j, k = −4
−jπ/3
e , k = −3



ak = e jπ/3
, k=3

2/j,

 k =4


0, otherwise for −7 ≤ k ≤ 8
Hence the DTFT over one period is
   
4π  π −jπ/3 3π jπ/3 3π 4π  π
X(Ω) = − δ Ω + +2πe δ Ω+ +2πe δ Ω− + δ Ω−
j 2 8 8 j 2
for π < Ω ≤ π. 

7.3.4 Properties of the DTFT


The properties of the discrete-time Fourier transform closely parallel those of
the continuous-time transform. In this section, we consider some of the more
useful properties.

Periodicity
We saw that the discrete-time Fourier transform is periodic in Ω with period
2π, so that
X(Ω + 2π) = X(Ω)

Linearity
Let x1 [n] and x2 [n] be two sequences with Fourier transforms X1 (Ω) and X2 (Ω),
respectively. Then
F[α1 x1 [n] + α2 x2 [n]] = α1 X1 (Ω) + α2 X2 (Ω)
10 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

Time and Frequency Shifting


It can be easily ahown that If
F
x[n] ←→ X(Ω)

then
F
x[n − n0 ] ←→ e−jΩn0 X(Ω)
and
F
ejΩ0 n x[n] ←→ X(Ω − Ω0 )

Differentiation in Frequency
dX(Ω)
F
nx[n] ←→ j
dΩ
Example 7.9 Find the discrete-time Fourier transform for the signal x[n] = nαn u[n], with
|α| < 1.

 Solution Then, by using the results in Example 5.4, we can write

d d 1
X(Ω) = j F [αn u[n]] = j
dΩ dΩ 1 − αe−jΩ
αe−jΩ
= 2 
(1 − αe−jΩ )

Convolution
F
x[n] ∗ h[n] ←→ X(Ω)H(Ω)

Multiplication
F 1
x[n].h[n] ←→ X(Ω) ∗ H(Ω)

Parseval’s Theorem
∞ Z
X
21
[x[n]] = |X(Ω)|2 dΩ
n=−∞
2π 2π

7.4 Relationship between CTFT and DTFT


The sampling theorem was originally presented in Chapter 6. We showed that
sampling is equivalent to amplitude modulation of a periodic impulse train. The
signal being sampled, x(t), multiplies the impulse train δT (t) to produce a new
signal xs (t),
xs (t) = x(t)δT (t)
The periodic impulse train samples the signal x(t) at the times at which the
impulses occur. The resulting continuous-time signal xs (t) is an impulse train
where the strengths of the impulses are the values of x(t) at the times equal
7.4. RELATIONSHIP BETWEEN CTFT AND DTFT 11

xs (t) = x(t)δT (t)

Figure 7.5: (a) Signal x(t) (solid line) and modulated impulse train xs (t), and (b)
sequence of samples x[n] = x(nT ).

to nT . This is shown in Figure 7.5(a). If we extracted the sample values from


the impulses and arranged them in a sequence, we would have the discrete-time
sequence of samples x[n] = x(nT ), shown in Figure 7.5(b).

In Section 6.3 we derived a frequency-domain representation of the sampled


signal xs (t) using the multiplication property as


1 X
Xs (ω) = X(ω − nωs ) (7.7)
T n=−∞

where ωs = 2π/T . An alternative form to X(ω) in (7.7) could be obtained using


the shifting property of Fourier transforms. The sampled signal xs (t) can be
written as

xs (t) = x(t)δT (t)


X∞
= x(t) δ(t − nT )
n=−∞

X
= x(t)δ(t − nT )
n=−∞
X∞
= x(nT )δ(t − nT )
n=−∞
12 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

Using the shifting property the Fourier transform of the signal xs (t) is

∞ ∞
F
X X
xs (t) = x(nT )δ(t − nT ) ←→ Xs (ω) = x(nT )e−jωnT
n=−∞ n=−∞

which can be written as



X
Xs (ω) = x[n]e−jωnT (7.8)
n=−∞

Clearly the frequency-domain represention Xs (ω) in (7.7) should be equivalent


to the one in (7.8), that is

∞ ∞
X 1 X
x[n]e−jωnT = X(ω − nωs )
n=−∞
T n=−∞

Using the definition of the DTFT in (7.6), we can state the following fundamen-
tal relation between the DTFT of a sequence of samples x[n] = x(nT ) and the
continuous-time Fourier transform of x(t) in sampling:


X(Ω) = Xs (ω)

ω= T

Thus, X(Ω) can be obtained from Xs (ω) by replacing ω with Ω/T . Therefore,
X(Ω) is identical to Xs (ω) frequency scaled by a factor T , as shown in Figure
7.6.

x(t) X(ω)

xs (t) Xs (ω)

x[n] X(Ω)

Figure 7.6: Connection between DTFT and CTFT


7.5. THE DISCRETE FOURIER TRANSFORM 13

7.5 The Discrete Fourier Transform


In order to motivate the study of the DFT, let us assume we are interested
in finding the Fourier transform of a continuous-time signal x(t) using a digital
computer. Since a digital computer can only store and manipulate a finite set of
numbers, it is necessary to represent x(t) by a finite set of values. The first step
in doing so is to sample the signal to obtain a discrete sequence xs [n]. Since the
signal may not be time-limited, the next step is to obtain a finite set of samples
of the discrete sequence by a process of truncation. Without loss of generality,
we can assume that these samples are defined for n in the range [0, N − 1]. Let
us denote this finite sequence by x[n].

Since we now have a discrete sequence, we can take the discrete-time Fourier
transform of the sequence as
N
X −1
X(Ω) = x[n]e−jΩn
n=0

This in still not in a form suitable for digital computation, sinc Ω is a continuous
variable taking values in the range [0, 2π]. Therefore, we need to evaluate X(Ω)
at only a finite number of values, Ωk , by a process of sampling uniformly in the
range [0, 2π] as
N
X −1
X(Ωk ) = x[n]e−jΩk n , k = 0, 1, . . . , M − 1 (7.9)
n=0

where

Ωk =k
M
The number of frequency samples, M , can be of any value. However, we choose
it to be the same as the number of time samples, N . With this modification,
and writing X(Ωk ) as X[k], we finally have
N −1
X 2π
X[k] = x[n]e−j N nk (7.10)
n=0

Equation (7.10) is known as the Discrete Fourier Transform. We saw that


X(Ω) is periodic in Ω with period 2π, so that X(Ωk ) = X(Ωk + 2π). This can
be written as
 

X[k] = X(Ωk ) = X(Ωk + 2π) = X (k + N ) = X[k + N ]
N

That is, X[k] is periodic with period N . Often, the length-N DFT sequence is
referred to as the N -point DFT.

The inverse discrete Fourier transform (IDFT) relation is given by


N −1
1 X 2π
x[n] = X[k]ej N nk (7.11)
N n=0
14 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

It can be shown that x[n] as determined from this equation is also periodic with
period N ,

N −1
1 X 2π
x[n + N ] = X[k]ej N (n+N )k
N n=0
N −1
1 X 2π
= X[k]ej N nk
N n=0
= x[n]

That is, the IDFT operation yields a periodic sequence, of which only the first
N values, corresponding to one period, are evaluated. Thus, in all operations
involving DFT and the IDFT, we are effectively replacing finite sequence x[n]
by its periodic extension, xp [n], as shown in Figure 7.7. One would expect there
is a connection between the Fourier series expansion of periodic discrete-time
sequences discussed earlier and the DFT. A connection explored next.

DFT and DTFS, Any Connection?


Comparision of Equations (7.4) and (7.5) with Equations (7.10) and (7.11) shows
that DFT X[k] of finite sequence x[n] can be interpreted as the coefficients ak
in the Fourier series representation of its periodic extension multiplied by the
period N . In fact the two can be made identical by including the factor 1/N
with the DFT rather than with the IDFT. Many authors and textbooks consider
both to be the same.

The different forms of the Fourier representation of discrete-time functions are


closely related to sampling. We saw earlier as shown in Figure 7.6, time-sampling
a function produces a periodic specturm. On the other hand, periodic repeti-
tions of a signal amounts to sampling its spectrum as depicted in Figure 7.7.
The dual of the time-sampling is the frequency-sampling.

Example 7.10 Suppose that x[n] = 1, 2, 2, 1, and x[n] = 0 for all other n. Compute the DFT
of x[n].

X(Ω)
x[n]

n Ω

X(Ω)
xp [n] Ωk

N n Ω
N

Figure 7.7: Periodic repetition of a signal results in sampling and periodic repetiton
of its spectrum.
7.5. THE DISCRETE FOURIER TRANSFORM 15

 Solution With N = 4 (i.e., 4-point DFT), from (7.10)

3
X π
X[k] = = x[n]e−j 2 nk , k = 0, 1, 2, 3
n=0

= x[0] + x[1]e−jπk/2 + x[2]e−jπk + x[3]e−jπ3k/2 , k = 0, 1, 2, 3


−jπk/2 −jπk −jπ3k/2
= 1 + 2e + 2e +e , k = 0, 1, 2, 3

Hence,

X[0] = 1 + 2 + 2 + 1 = 6

X[1] = 1 + 2e−jπ/2 + 2e−jπ + e−jπ3/2


= 1 − j2 − 2 + j = −1 − j

X[2] = 1 + 2e−jπ + 2e−j2π + e−j3π


=1−2+2−1=0

X[3] = 1 + 2e−jπ3/2 + 2e−j3π + e−jπ9/2


= 1 + j2 − 2 − j = −1 + j 

Again consider the signal in Example 7.10 with the 4-point DFT given by Example 7.11


 6, k=0

−1 − j, k = 1
X[k] =


0, k=2
−1 + j, k = 3

Compute the Inverse DFT to determine x[n].

 Solution From (7.11)

3
1X π
x[n] = X[k]ej 2 nk
4 n=0
1h i
= X[0] + X[1]ejπn/2 + X[2]ejπn + X[3]ejπ3n/2
4

Thus,

1
x[0] = [X[0] + X[1] + X[2] + X[3]] = 1
4
1 1
x[1] = [X[0] + jX[1] − X[2] − jX[3]] = [8] = 2
4 4
1
x[2] = [X[0] − X[1] + X[2] − X[3]] = 2
4
1 1
x[3] = [X[0] − jX[1] − X[2] + jX[3]] = [4] = 1 
4 4
16 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS

7.6 Examples of the Use of The DFT


The primary application of the DFT is to approximate the Fourier transform of
signals. To demonstate the use of DFT, consider the function
 
t−1
x(t) = rect
2

We choose a sampling frequency of ωs = 10π rad/sec or 5 Hz to obtain a discrete-


time sequence of samples x[n] a shown in Figure 7.8(a). We choose to use 16
samples for the computation. The 16-point DFT is shown in Figure 7.8(b).
Because
the DFT is calculated from discrete-time samples, we must multiply
X[k] by the sampling period T to cancel ot the factor of 1/T inherent to the
Fourier transform of sampled signals.

x[n]


X[k]

Figure 7.8: A sampled rectangular pulse and its 16-point DFT.


7.6. EXAMPLES OF THE USE OF THE DFT 17

Figure 7.9 shows a comparision of the Fourier transform approximation found


by computing the DFT and the actual Fourier transform.

Figure 7.9: Comparision of the DFT with the Fourier transform.


18 CHAPTER 7. DISCRETE TIME SIGNALS AND SYSTEMS
Bibliography

[1] Roberts, M. J., Signals and Systems: Analysis using Transform Meth-
ods and MATLAB R
. International edition, McGraw-Hill 2003.
[2] Haykin, S., and Van Veen, B., Signals and Systems, New York, John
Wiley & Sons, 1999.

[3] Philips, C., Parr, J., and Riskin, E., Signals, Systems, and Transforms,
Upper Saddle River, NJ, Prentice-Hall, 2008.
[4] Stremler, F., Introduction to Communication Systems, Reading, MA
Addison-Wesley 1990.

19

You might also like