1 of 133

Module 4�Fourier Representation of aperiodic signals

Syllabus: Definitions, basic problems and Properties with examples

Krishna Teja

2 of 133

Discrete-Time Nonperiodic Signals: The Discrete-Time Fourier Transform

  • We develop DTFT from DTFS by describing nonperiodic signal as

the limit of a periodic signal whose period N approaches infinity.

  • We assume nonperiodic signal is represented by single period of

periodic signal that is centered on origin and the limit as N

approaches infinity is taken in a symmetric manner.

  • Let

be a periodic signal with period N=2M+1

  • Define the nonperiodic signal as one period of

Krishna Teja

3 of 133

Fig 1: Approximation of nonperiodic signal with periodic signal (a) Nonperiodic signal

(b) Periodic approximation

Krishna Teja

4 of 133

  • Note that as M increases, the periodic replicates of x[n] move

farther and farther away from the origin.

  • Begin with DTFS representation of periodic signal. We have DTFS

pair as

---------------------->(1)

---------------------->(2)

Since

we may rewrite (2) in terms of x[n]

Krishna Teja

5 of 133

  • We now define a continuous function of frequency

whose samples at

are DTFS coefficients normalized by 2M+1

---------------------->(3)

so that

Substitute this definition of X[k] into (1)

Using the relationship

Krishna Teja

6 of 133

Consider the effect as

on fundamental frequency

  • As M increases,

decreases and the spacing between harmonics

in DTFS decreases.

Note that

is 2π periodic in

---------------------->(4)

Krishna Teja

7 of 133

Fig 2: Example spectra

for increasing value of M.

M increases from (a) to (c) so

decreases.

Substituting the limiting condition in (4), we get

Krishna Teja

8 of 133

We are summing the values of a function

evaluated at

multiplied by the width between the samples

This is the rectangular rule approximation to an integral.

Taking the limits and identifying

and

---------------------->(5)

---------------------->(6)

We have thus expressed x[n] as a weighted superposition of discrete-time sinusoids.

In this case superposition is integral and weighting on each sinusoid is

Krishna Teja

9 of 133

DTFT Representation

The DTFT representation is expressed as

where

We say that

and

are a DTFT pair and write

The transform

describes the signal

as a function of

sinusoidal frequency Ω

and is termed as frequency domain representation of

---------------------->(1)

---------------------->(2)

Krishna Teja

10 of 133

  • The DTFT is primarily used to analyse the action of discrete-time systems on discrete-time signals.

 

 

 

 

 

 

 

It can be shown that sum in (1) converges in a mean-square sense but does not converge pointwise.

 

Krishna Teja

11 of 133

Ex 1: Find the DTFT of the sequence

Sol:

We have

This sum diverges for

For

we have convergent geometric series:

Recall:

Krishna Teja

12 of 133

If

is real valued, we may write magnitude and phase of

The magnitude and phase are depicted graphically for

Note that both are periodic with 2π

Krishna Teja

13 of 133

Fig 1: DTFT of x[n] (a) Magnitude and (b) Phase Spectra

Krishna Teja

14 of 133

 

Sol:

 

 

We have the DTFT as

 

 

 

 

 

 

Krishna Teja

15 of 133

 

 

 

 

Recall:

 

 

 

Krishna Teja

16 of 133

3. Let

 

 

 

 

 

 

 

 

 

 

Sol:

We have the DTFT as

 

 

Change of variable

 

 

 

DTFT of rectangular pulse

Krishna Teja

17 of 133

 

 

Recall:

 

 

 

---------------------->(1)

 

 

Krishna Teja

18 of 133

 

 

 

 

 

From (1)

 

 

L’Hospital’s rule gives

 

 

Krishna Teja

19 of 133

 

 

 

 

 

 

 

 

 

 

 

 

Zero crossings,

 

 

 

 

 

 

Fig 2: DTFT in frequency domain

Krishna Teja

20 of 133

Inverse DTFT of a rectangular spectrum

4. Find the inverse DTFT of

 

which is depicted in Fig 1.

 

 

 

 

 

 

 

Fig 1: One period of rectangular pulse in frequency domain

Sol:

 

 

Krishna Teja

21 of 133

 

 

 

 

 

 

 

---------------------->(1)

 

 

 

 

Using L’Hospital’s, we can show that

 

 

 

We can write

Krishna Teja

22 of 133

We know that

 

 

 

 

 

Fig 2: Inverse DTFT in the time domain

Zero crossings,

 

 

 

Krishna Teja

23 of 133

 

DTFT of the Unit Impulse

Sol:

We have the DTFT as

 

 

 

 

 

 

DTFT

 

 

 

 

 

 

 

 

 

 

 

Fig 2: DTFT of unit impulse in frequency domain

Fig 1: Unit impulse in time domain

Krishna Teja

24 of 133

Inverse DTFT of the Unit Impulse Spectrum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig 1: Unit impulse in frequency domain

Sol:

 

 

 

Krishna Teja

25 of 133

 

 

 

 

 

 

 

DTFT

 

 

Fig 2: Inverse DTFT in time domain

 

Krishna Teja

26 of 133

Ex 7: Find the DTFT of the following time domain signals.

 

 

Sol:

We have the DTFT as

 

 

 

Recall:

 

 

 

 

Krishna Teja

27 of 133

 

 

Sol:

We have the DTFT as

 

 

 

 

 

 

 

 

In first summation,

 

 

 

 

 

 

 

 

 

Recall:

Krishna Teja

28 of 133

 

 

 

 

 

 

 

 

Krishna Teja

29 of 133

 

 

We have the DTFT as

 

 

 

 

 

 

 

 

 

 

Sol:

Krishna Teja

30 of 133

Ex 8: Find the inverse DTFT of the following frequency-domain signals:

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

---------------------->(1)

Krishna Teja

31 of 133

 

 

 

 

 

Krishna Teja

32 of 133

 

 

 

 

 

 

Sol:

 

 

We have the inverse DTFT calculated as

 

 

 

 

 

 

---------------------->(1)

Krishna Teja

33 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

34 of 133

 

 

Fig 1

 

 

 

 

 

We have the inverse DTFT calculated as

 

 

Sol:

Krishna Teja

35 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

36 of 133

 

 

 

 

 

Krishna Teja

37 of 133

Continuous-Time Nonperiodic Signals: The Fourier Transform

 

 

  • Thus the FT representation of the signal involves the integral over the entire frequency interval; that is

 

where

 

 

 

 

Krishna Teja

38 of 133

 

 

 

 

FT

 

 

Krishna Teja

39 of 133

 

 

 

 

  • The size of each discontinuity is finite.

Krishna Teja

40 of 133

FT of a Real Decaying Exponential

 

 

 

 

 

Sol:

 

 

 

 

 

 

 

 

Fig 1: Real time domain exponential signal

Krishna Teja

41 of 133

 

 

 

 

Magnitude:

 

 

 

 

 

 

 

 

Fig 2: Magnitude spectrum

 

 

 

Krishna Teja

42 of 133

 

 

 

 

 

 

 

 

 

 

 

Phase:

Fig 2: Phase spectrum

Krishna Teja

43 of 133

FT of a Rectangular Pulse

Ex 10: Consider a rectangular pulse depicted in Fig 10 and defined as

 

 

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig 1: Rectangular pulse in time domain

Krishna Teja

44 of 133

 

 

 

 

 

Zero crossings,

 

 

 

 

 

 

 

 

Fig 1: FT in frequency domain

Using sinc notation,

 

 

 

Krishna Teja

45 of 133

 

 

 

FT

 

FT pair

Krishna Teja

46 of 133

Ex 11: Find the FT of the following signals:

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

47 of 133

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

48 of 133

 

 

 

 

 

 

 

FT

Krishna Teja

49 of 133

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

50 of 133

 

 

 

 

 

 

 

 

 

 

 

Sol:

 

 

 

 

 

 

 

 

 

 

Krishna Teja

51 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

52 of 133

Inverse FT of a Rectangular Spectrum

Ex 12. Find the inverse FT of

 

which is depicted in Fig 1.

 

 

 

 

 

Fig 1: Rectangular spectrum in the frequency domain

Sol: The inverse FT gives

 

 

 

---------------------->(1)

Krishna Teja

53 of 133

 

 

 

 

 

 

 

Hence we write

 

 

 

 

 

 

 

 

 

 

Fig 1: Inverse FT in time domain

Zero crossings,

 

 

 

 

 

FT

Krishna Teja

54 of 133

FT of the Unit Impulse

 

 

 

 

 

 

 

 

FT

 

Krishna Teja

55 of 133

Inverse FT of an Impulse Spectrum

 

Sol:

 

The inverse FT gives

 

 

 

 

 

 

 

FT

 

Krishna Teja

56 of 133

Ex 15: Find the inverse FT of the following spectra:

 

 

 

 

Sol:

The inverse FT is given as

 

 

 

 

 

 

 

 

Krishna Teja

57 of 133

 

 

 

 

 

 

Krishna Teja

58 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig 15: Frequency domain signal

Krishna Teja

59 of 133

The inverse FT is given as

 

 

 

 

 

 

 

 

 

Krishna Teja

60 of 133

 

 

 

 

 

 

Krishna Teja

61 of 133

 

 

Fig 16: Frequency domain signal

 

 

 

 

 

 

 

 

The inverse FT is given as

 

 

 

 

 

 

 

Krishna Teja

62 of 133

 

 

 

 

Krishna Teja

63 of 133

The Four Fourier Representations

Time Domain

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Krishna Teja

64 of 133

  • The frequency domain representation of nonperiodic signals is a continuous function of frequency.
  • Signals that are periodic in time have discrete frequency domain representations, while nonperiodic time signals have continuous frequency domain representations.

Time-Domain

property

Frequency-Domain

property

continuous

nonperiodic

discrete

periodic

periodic

discrete

nonperiodic

continuous

Table: Periodic properties of Fourier Representations

Krishna Teja

65 of 133

Linearity and Symmetry Properties

If

 

 

 

and

 

 

 

then

 

 

 

1a. Linearity property

Proof:

 

 

 

 

 

 

 

 

Hence

Krishna Teja

66 of 133

Similarly

 

 

 

Ex 16: Use the linearity property to determine the Fourier representation of the following

signals:

 

 

Sol:

 

Recall

 

 

 

 

 

 

 

 

 

Krishna Teja

67 of 133

 

 

Sol:

Recall the DTFT pair

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hence

 

 

 

Krishna Teja

68 of 133

1b. Symmetry properties: Real and Imaginary signals

 

 

 

 

 

---------------------->(1)

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

69 of 133

 

 

 

Substituting in (1), we get

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

70 of 133

1b. Symmetry properties: Even and Odd signals

 

 

 

Substituting in (1), we get

 

 

 

 

 

 

 

 

 

 

Krishna Teja

71 of 133

 

 

 

Substituting in (1), we get

 

 

 

 

 

 

 

 

 

 

Krishna Teja

72 of 133

Symmetry properties of DTFT

 

 

 

 

 

 

 

 

Krishna Teja

73 of 133

Ex 17: Determine whether the time domain signals corresponding to the following frequency-domain representations are real or complex valued and even or odd:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sol:

 

 

 

 

 

 

 

 

Krishna Teja

74 of 133

 

 

Sol:

 

 

 

 

 

 

 

 

Hence

 

 

 

Sol:

 

 

 

Hence

 

 

Krishna Teja

75 of 133

3. Convolution property

  • Convolution of signals in time domain transforms to multiplication of their respective Fourier representations in frequency domain.
  • With convolution property, we may analyse the input-output behaviour of linear system in frequency domain by multiplying the transforms instead of convolving time signals.

 

 

 

---------------------->(1)

 

 

 

 

 

 

---------------------->(2)

Substitute (2) in (1), we get

Krishna Teja

76 of 133

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

77 of 133

Solving a Convolution problem in Frequency Domain

 

Sol:

This problem is extremely difficult to solve in the time domain

Recall the FT pair

 

 

FT

Using the convolution property,

 

 

 

 

FT

 

 

FT

 

Krishna Teja

78 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

We conclude that

 

Krishna Teja

79 of 133

Finding Inverse FT’s by means of convolution property

Ex 19:

 

 

FT

 

Sol:

 

Recall the FT pair

 

 

 

FT

 

 

where

 

Using the convolution property,

 

 

 

Krishna Teja

80 of 133

 

 

 

 

 

 

FT

 

where

Hence

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

81 of 133

 

 

 

 

 

 

 

A similar property holds for convolution of discrete-time nonperiodic signals

If

 

 

 

and

 

 

 

then

 

 

 

Krishna Teja

82 of 133

Ex 20:

 

 

 

Sol:

Recall the DTFT pair

 

 

 

 

 

 

Using the convolution property,

 

 

 

 

 

 

Hence

 

 

 

 

Krishna Teja

83 of 133

 

 

Sol:

 

 

 

Ex 21:

Use the convolution property to find the time-domain signals corresponding to the following frequency domain representations:

 

 

Sol:

Recall the FT pair

 

 

 

 

 

 

FT

 

 

 

 

 

Krishna Teja

84 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

85 of 133

 

 

Sol:

Recall the DTFT pair

 

 

 

Using the convolution property,

 

 

 

 

 

 

---------------> (1)

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

86 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

87 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

88 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Krishna Teja

89 of 133

 

 

 

 

 

 

 

 

 

Krishna Teja

90 of 133

3. Differentiation property

 

 

 

 

 

 

 

 

 

From which it follows that

 

Krishna Teja

91 of 133

Ex 22:

Verifying the Differentiation property

The differentiation property implies that

 

 

 

Verify this result by differentiating and taking FT of the result.

Sol:

Using the product rule for differentiating, we have

 

 

 

 

 

 

 

 

 

Krishna Teja

92 of 133

Ex 23: Use the differentiation property to find the FT of the following signals:

 

 

Sol:

 

 

FT

Recall the FT pair

 

 

Take FT on both sides of this equation

 

 

 

 

FT

 

Krishna Teja

93 of 133

 

 

Sol:

Let

 

 

---------------> (1)

 

 

 

 

 

 

 

Take FT on both sides

 

 

Krishna Teja

94 of 133

 

 

 

 

 

Sol:

Let

 

 

 

 

---------------> (1)

Recall the FT pair

 

 

FT

 

 

Using the differentiation property on (1), we get

 

 

 

 

Krishna Teja

95 of 133

Differentiation in frequency

 

 

 

 

 

 

 

From which it follows that

 

Krishna Teja

96 of 133

Ex 24:

 

Sol:

 

 

 

 

 

 

 

 

The differentiation in time property states

 

 

 

 

 

 

The differentiation in frequency property states

 

 

 

 

 

 

---------------> (1)

---------------> (2)

Krishna Teja

97 of 133

Since lhs of (1) and (2) are equal, the rhs must also be equal

 

 

 

 

 

 

 

 

 

Krishna Teja

98 of 133

Ex 25:

Use the frequency-differentiation property to find the FT of

 

Sol:

The differentiation in frequency property states

 

 

 

Let

 

 

Hence

 

 

 

 

 

Krishna Teja

99 of 133

Ex 26:

Use the time-differentiation and convolution properties to find the FT of

 

Sol:

Let us define the signals

 

 

Use the FT pair

 

 

 

 

 

 

Let

 

 

---------------> (1)

---------------> (2)

Take FT on equations (1) and (2), we get

 

 

 

 

Krishna Teja

100 of 133

  • The operation of differentiation does not apply to discrete-valued quantities and thus a frequency differentiation property for the FS and DTFS does not exist.
  • A frequency differentiation property does exist for DTFT

 

 

 

 

 

Krishna Teja

101 of 133

Ex 27:

Use the frequency-differentiation property to find the DTFT of

 

Sol:

 

 

 

 

 

 

 

Krishna Teja

102 of 133

Ex 28:

 

 

Sol:

Recall the DTFT pair

 

 

 

We are given the DTFT, let us define

 

 

 

Taking Inverse FT using frequency differentiation property, we get

 

 

 

Krishna Teja

103 of 133

4. Integration property

  • The operation of integration applies only to continuous dependent variables.
  • We may integrate with respect to time in both the FT and FS and with respect to frequency in the FT and DTFT.
  • We define

 

 

 

Using differentiation property,

 

---------------> (1)

---------------> (2)

 

Krishna Teja

104 of 133

 

  • We desire to apply the integration property to signals that do not have a zero time averaged value.

 

  • We get around this problem by including impulses in the transform.

 

 

 

 

 

Krishna Teja

105 of 133

 

Sol:

The unit step can be expressed as integration of the unit impulse.

 

Since

 

 

 

Using the integration property,

 

 

 

 

 

Krishna Teja

106 of 133

Here, the signum function is defined as

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Using the standard results, we get

 

 

FT

Krishna Teja

107 of 133

 

 

 

FT

 

Taking the FT on both the sides, we get

 

 

 

 

 

 

Krishna Teja

108 of 133

Ex 30:

Use the integration property to find the inverse FT of

 

Sol:

We have the integration property defined as

 

 

 

Therefore

where

 

By taking inverse FT, we get

 

 

 

 

 

Krishna Teja

109 of 133

4. Time Shift property

 

 

 

 

 

 

 

 

 

 

 

Similarly

 

 

 

Krishna Teja

110 of 133

 

 

 

 

 

 

 

 

 

 

 

 

 

Sol:

From the fig, we show that

 

Using time shift property,

 

 

 

 

FT

 

FT pair

 

Krishna Teja

111 of 133

Ex 32: Find the FT of the following time domain signals:

 

Sol:

Recall the DTFT pair

 

 

 

Let

 

 

Using time shift property,

 

 

 

 

 

 

 

Krishna Teja

112 of 133

 

Sol:

 

 

Let

 

 

 

---------------> (1)

---------------> (2)

---------------> (3)

Apply FT on equations (1), (2) and (3)

 

 

 

 

 

 

Krishna Teja

113 of 133

Ex 33:

Find the time domain signals corresponding to the following Fourier representations:

 

Sol:

Recall the FT pair

 

 

 

Let

 

 

Apply inverse FT on the above equation using time shift property

 

where

 

 

Krishna Teja

114 of 133

4. Frequency Shift property

Suppose

 

 

FT

 

Let

 

 

FT

 

 

 

 

 

 

 

 

 

 

A frequency shift corresponds to multiplication in the time domain by a complex sinusoid whose frequency is equal to the shift.

Krishna Teja

115 of 133

Similarly

 

 

 

Ex 34:

Use the frequency shift property to determine the FT of the complex sinusoidal pulse

 

 

 

Sol:

 

 

 

FT

 

FT pair

Let

 

 

 

 

 

Krishna Teja

116 of 133

 

 

 

 

Ex 35:

Use the frequency shift property to determine the time domain signals corresponding to the following Fourier representations:

 

Sol:

Let

 

 

 

 

 

Using the frequency shift property

 

Krishna Teja

117 of 133

 

Sol:

 

Taking the inverse FT using the frequency shift property, we get

 

 

 

 

where

Taking the inverse FT

 

 

 

 

 

Krishna Teja

118 of 133

Ex 36:

Find the FT of the signal

 

Sol:

To solve this problem, we require three properties

  • Differentiation in time
  • Convolution
  • Time shifting

Let

 

 

 

Applying convolution and differentiation property,

 

The transform pair

 

 

 

 

 

Krishna Teja

119 of 133

 

 

 

Krishna Teja

120 of 133

5. Scaling properties

Consider the effect of scaling the time variable on the frequency domain representation of a signal.

 

 

 

 

 

 

 

Krishna Teja

121 of 133

These two integrals may be combined into single integral

 

 

 

 

 

 

Hence scaling in time introduces the inverse scaling in the frequency domain representation and an amplitude change.

Krishna Teja

122 of 133

 

Krishna Teja

123 of 133

Ex 37:

Let the rectangular pulse

 

 

 

 

 

 

 

Sol:

Recall the FT pair

 

 

 

FT

 

 

 

Note that

 

 

Using the time scaling property,

Krishna Teja

124 of 133

 

 

 

 

 

 

Krishna Teja

125 of 133

6. Parseval Relationships

  • The Parseval relationship states that the energy or power in the time domain representation of a signal is equal to the energy or power in the frequency-domain representation.
  • The energy or power is conserved in the Fourier representation.
  • The energy in the continuous-time nonperiodic signal is

 

 

 

 

 

---------------> (1)

---------------> (2)

Substituting (2) in (1) we get,

Krishna Teja

126 of 133

 

 

We interchange the order of integrations,

 

 

 

so we conclude that

 

 

 

Krishna Teja

127 of 133

Ex 38:

Let

 

 

Sol:

Recall the DTFT pair

 

 

 

According to Parseval’s theorem

 

 

 

 

 

Krishna Teja

128 of 133

Ex 39:

Use Parseval’s theorem to evaluate the quantity

 

Sol:

 

 

We have the Parseval’s theorem stated as

 

 

We can find that

 

 

 

 

Krishna Teja

129 of 133

7. Duality

  • We observed a consistent symmetry in time and frequency domain representations of signals.
  • These symmetries are consequence of symmetry I definitions of time and frequency domain representations.
  • This interchangeability is termed as duality.

 

and

 

 

Krishna Teja

130 of 133

 

 

 

 

 

FT

 

 

 

 

FT

Krishna Teja

131 of 133

These relationships imply a certain symmetry between the role of time and frequency. If we are given a FT pair

 

 

FT

then we interchange the role of time and frequency to obtain the new FT pair

 

 

FT

 

Krishna Teja

132 of 133

Fig: Duality of rectangular pulses and sinc functions

Krishna Teja

133 of 133

Ex 40:

Find the FT of

 

Sol:

First we recognize that

 

 

 

 

 

 

 

FT

This implies that

 

 

 

Krishna Teja