Module 4

The Fourier Transform

Winter 2024

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 1

 Agenda
• Review: Fourier Series of Periodic Signals
• Aperiodic Signal Representation by the Fourier Integral
• Fourier Transform of Aperiodic Signals
• Transforms of Some Useful Functions
• Properties of the Fourier Transform

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 2

 Table 1 Fourier Series Representation of a Periodic Signal of Period 𝑇0 (𝜔0 = 2𝜋ൗ𝑇0)
Series Form Coefficient Computation Conversion Formulas
Trigonometric 1
∞ 𝑎0 = න 𝑥 𝑡 𝑑𝑡 𝑎0 = 𝐶0 = 𝐷0
𝑇0
𝑓 𝑡 = 𝑎0 + ෍ (𝑎𝑛 cos 𝑛𝜔0 𝑡 + 𝑏𝑛 sin 𝑛𝜔0 𝑡)
𝑛=1 2 𝑎𝑛 − 𝑗𝑏𝑛 = 𝐶𝑛 𝑒 𝑗𝜃𝑛 = 2𝐷𝑛
𝑎𝑛 = න 𝑥 𝑡 cos 𝑛𝜔0 𝑡 𝑑𝑡
𝑇0
𝑎𝑛 + 𝑗𝑏𝑛 = 𝐶𝑛 𝑒 −𝑗𝜃𝑛 = 2𝐷−𝑛
2
𝑏𝑛 = න 𝑥 𝑡 sin 𝑛𝜔0 𝑡 𝑑𝑡
𝑇0

Compact Trigonometric 𝐶0 = 𝑎0
∞ 𝐶0 = 𝐷0
𝑓 𝑡 = 𝐶0 + ෍ 𝐶𝑛 cos(𝑛𝜔0 𝑡 + 𝜃𝑛 )
𝑛=1 𝐶𝑛 = 𝑎𝑛2 + 𝑏𝑛2 𝐶𝑛 = 2 𝐷𝑛 𝑛 ≥ 1

𝜃𝑛 = ∠𝐷𝑛
−𝑏𝑛
𝜃𝑛 = tan−1
𝑎𝑛

Exponential 1 𝑇0/2
∞ 𝐷𝑛 = න 𝑓(𝑡)𝑒 −𝑗𝑛𝜔0 𝑡 𝑑𝑡 𝐶𝑛
𝑗𝑛𝜔0 𝑡
𝑇0 −𝑇0/2 𝐷𝑛 = 𝐷−𝑛 =
𝑓 𝑡 = ෍ 𝐷𝑛 𝑒 2

𝑛=−∞
∠𝐷𝑛 =𝜃𝑛 and ∠𝐷−𝑛 =−𝜃𝑛

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 3

 Existence of the Fourier Series
Dirichlet Conditions
• The function x(t) must be absolutely interable:
𝑇0 /2
න |𝑥 𝑡 |𝑑𝑡 < ∞
−𝑇0 /2

• The function x(t) must have only finite number of finite

discontinuities in one period.

• The function x(t) must contain only a finite number of maxima and
minima in one period.

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 4

 Construction of a periodic signal

The period 𝑇0 is made

long enough to avoid
overlap between pulse.

If we let 𝑇0 → ∞,
the pulses in the periodic signal repeat
after an infinite interval.

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 5

If we let 𝑇0 → ∞,
the pulses in the periodic signal repeat after an infinite interval,
and therefore:

lim 𝑥𝑇0 𝑡 = 𝑥(𝑡)

𝑇0 →∞

Thus, the Fourier series representing 𝑥𝑇0 𝑡 will also represent

𝑥(𝑡) in the limit 𝑇0 → ∞.
∞

𝑥𝑇0 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0 𝑡
𝑛=−∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 6

Fourier Series Representation of Aperiodic Signal
when 𝑇0 → ∞

𝑥𝑇0 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0 𝑡
𝑛=−∞ 𝑆𝑖𝑛𝑐𝑒 𝑇0 → ∞

1 𝑇0/2 1 ∞
𝐷𝑛 = න 𝑥 𝑡 𝑒 −𝑗𝑛𝜔0 𝑡 𝑑𝑡 = න 𝑥(𝑡)𝑒 −𝑗𝑛𝜔0 𝑡 𝑑𝑡
𝑇0 −𝑇0/2 𝑇0 −∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 7

 1 ∞
𝐷𝑛 = න 𝑥(𝑡)𝑒 −𝑗𝑛𝜔0 𝑡 𝑑𝑡
𝑇0 −∞

∞
Define: 𝑋 𝜔 = ‫׬‬−∞ 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡

Then: 𝑇0 𝐷𝑛 = 𝑋(𝜔)|𝜔=𝑛𝜔0

𝑋(𝜔) is the envelope of 𝑇0 𝐷𝑛 .

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 8

 1 ∞ −𝑗𝑛𝜔0 𝑡 𝑑𝑡 1
𝐷𝑛 = ‫׬‬ 𝑥(𝑡)𝑒 = X(n𝜔0 )
𝑇0 −∞ 𝑇0

1
The Fourier coefficients 𝐷𝑛 are 𝑇 times the
0
samples of 𝑋 𝜔 uniformly spaced at intervals of
𝜔0 .

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 9

 𝑇0 𝐷𝑛 =𝑋(𝑛𝜔0 ) 1
𝑥𝑇0 𝑡 = σ∞ 𝐷
𝑛=−∞ 𝑛 𝑒 𝑗𝑛𝜔0 𝑡
𝑥𝑇0 𝑡 ∞
= σ𝑛=−∞ 𝑋(𝑛𝜔0 ) 𝑒 𝑗𝑛𝜔0 𝑡
𝑇 0

𝜔
𝑇0 = 2𝜋0
∞ 1 1
𝑥𝑇0 𝑡 = σ𝑛=−∞ 𝑋(𝑛𝜔0 ) 𝑒 𝑗𝑛𝜔0 𝑡 𝑥𝑇0 𝑡 = σ∞ 𝑋(𝑛𝜔0 ) 𝑒 𝑗𝑛𝜔0 𝑡 𝜔0
𝑇 0 2𝜋 𝑛=−∞

As 𝑇0 → ∞, 𝜔0 → 0 𝑥𝑇0 𝑡 = 𝑥 𝑡
Fourier transform: represent a periodic signal with a particular at different frequencies

1 ∞
Or 𝜔0 → 𝑑𝜔 and ෍ → න ⇒ 𝑥(𝑡) = න 𝑋(𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔
2𝜋 −∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 10

 Fourier Transform
Fourier
𝒙(𝒕) 𝑿(𝝎) = ℱ[𝒙(𝒕)]
Transform

∞
𝑋 𝜔 = ℱ𝑥 𝑡 = න 𝑥 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡
−∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 11

 Fourier Transform
Fourier
𝒙(𝒕) 𝑿(𝝎) = ℱ[𝒙(𝒕)]
Transform

𝑋 𝜔 = 𝑅𝑒 𝜔 + 𝑗𝐼𝑚(𝜔) 𝐴 = 𝑋(𝜔) = 𝑅𝑒(𝜔)2 + 𝐼𝑚(𝜔)2

𝐼𝑚 𝜔
−1
∅ = ∠𝑋 𝜔 = − tan ( )
𝑅𝑒 𝜔

𝑋 𝜔 = 𝑋(𝜔) 𝑒 𝑗∠𝑋 𝜔 = 𝐴 sin(𝜔𝑡 + ∅)

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 12

Conjugate Symmetry Property of Fourier Transform

𝑥 𝑡 ⟺ 𝑋(𝜔)

𝑥 ∗ 𝑡 ⟺ 𝑋 ∗ (−𝜔)

If x(t) is real function of t:

𝑋 −𝜔 = 𝑋 ∗ (𝜔)

𝑋(−𝜔) = 𝑋(𝜔) ∠𝑋 −𝜔 = −∠𝑋 𝜔

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 13

 Inverse Fourier Transform
Inverse Fourier
𝑿(𝝎) 𝒙(𝒕) = ℱ −𝟏 [𝑿(𝝎)]
Transform

note that for the inverse is in frequency terms

∞
1
𝑥(𝑡) = ℱ −1 [𝑋(𝜔)] = න 𝑋(𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔
2𝜋 −∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 14

 Example 7.1 u(t) makes it a causal signal

Fourier Transform of a Casual Exponential 𝑒 −𝑎𝑡 𝑢 𝑡

By definitions:
∞ ∞ ∞
−1
𝑋 𝜔 = ℱ𝑥 𝑡 = න 𝑥 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 =න 𝑒 −𝑎𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 =න 𝑒 −(𝑎+𝑗𝜔)𝑡 𝑑𝑡 = 𝑒 −(𝑎+𝑗𝜔)𝑡 |∞
0
−∞ 0 0 𝑎 + 𝑗𝜔

If 𝑎 < 0, when 𝑡 → ∞, 𝑒 −(𝑎+𝑗𝜔)𝑡 = ∞. Therefore, 𝑋 𝜔 = ∞.

1
If a > 0, X ω =
𝑎+𝑗𝜔

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 15

 Example 7.1
Fourier Transform of a Casual Exponential 𝑒 −𝑎𝑡 𝑢 𝑡
We can express 𝑎 + 𝑗𝜔 in the polar form as:

𝜔
𝑗 tan−1 ( 𝑎 )
𝑎 + 𝑗𝜔 = 𝑎2 + 𝜔 2 𝑒

1 𝜔
−𝑗 tan−1 ( 𝑎 )
𝑋 𝜔 = 𝑒 = 𝑋(𝜔) 𝑒 𝑗∠𝑋 𝜔
𝑎2 + 𝜔 2

1 −1 𝜔
𝑋(𝜔) = ∠𝑋 𝜔 = − tan ( )
𝑎2 +𝜔2 𝑎

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 16

 Example 7.1
Fourier Transform of a Casual Exponential 𝑒 −𝑎𝑡 𝑢 𝑡
1 𝜔
−𝑗 tan−1 ( )
𝑋 𝜔 = 𝑒 𝑎 = 𝑋(𝜔) 𝑒 𝑗∠𝑋 𝜔
𝑎2 + 𝜔 2

1 𝜔
𝑋(𝜔) = ∠𝑋 𝜔 = − tan−1 ( )
𝑎2 +𝜔2 𝑎

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 17

 The Fourier Transform Existence

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 18

 Unit Gain Function
We define a unit gate function 𝑟𝑒𝑐𝑡 (𝑥) as a gate pulse of unit height
and width 𝜏:

𝑥
𝑥 𝑡 = 𝑟𝑒𝑐𝑡 ( )
𝜏

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 19

 Unit Triangle Function
We define a unit triangle function Δ(x) as triangular pulse of unit
height and width 𝜏:

𝑥
𝑥 𝑡 = ∆
𝜏

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 20

 Interpolation Function 𝑠𝑖𝑛𝑐(𝑡)
• The function sin x/x is the "sine over argument" function denoted by sinc
(x). This function plays an important role in signal processing. It is also
known as the filtering or interpolating function.

sin(𝜋𝑡)
𝑠𝑖𝑛𝑐 𝑡 =
𝜋𝑡

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 21

 Notes about 𝑠𝑖𝑛𝑐(𝑥) function
1. 𝑠𝑖𝑛𝑐 (𝑥) is an even function of 𝑥.
2. 𝑠𝑖𝑛𝑐 (𝑥) = 0 when sin 𝑥 = 0 except at 𝑥 = 0, where it appears to be
indeterminate. This means that 𝑠𝑖𝑛𝑐 𝑥 = 0 for 𝑥 = ±𝑝, ±2𝑝, ±3𝑝, … .
3. Using L'Hôpital's rule, we find 𝑠𝑖𝑛𝑐 (0) = 1.
4. 𝑠𝑖𝑛𝑐 (𝑥) is the product of an oscillating signal sin 𝑥 (of period 2π) and a
monotonically decreasing function 1/𝑥. Therefore, 𝑠𝑖𝑛𝑐 (𝑥) exhibits
damped oscillations of period 2π, with amplitude decreasing continuously as
1/𝑥.

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 22

 Example 7.2
Fourier Transform of a Rectangular Pulse
𝑡 ∞ 𝑡
ℱ 𝑟𝑒𝑐𝑡 = 𝑋 𝜔 = ‫׬‬−∞ 𝑟𝑒𝑐𝑡( ) 𝑒 −𝑗𝜔𝑡 dt
𝜏 𝜏

𝑡 𝜏
Since 𝑟𝑒𝑐𝑡 = 1 for 𝑡 < and zero otherwise,
𝜏 2 Euler's formula: sin(wt/2) = 1/2j * (e^jwt -e^-jwt)

𝜔𝜏
𝜏/2
1 −𝑗𝜔𝜏 2 sin( )
𝑋 𝜔 = න 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = − (𝑒 2 − 𝑒 𝑗𝜔𝜏/2 )= 2
−𝜏/2 𝑗𝜔 𝜔
Top and bottom by T 𝜔𝜏
sin( 2 ) 𝜔𝜏
𝑋 𝜔 = τ 𝜔𝜏 = τ𝑠𝑖𝑛𝑐( )
2
2

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 23

 Example 7.2
Fourier Transform of a Rectangular Pulse

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 24

 𝑡
Bandwidth of 𝑟𝑒𝑐𝑡
𝜏
• The spectrum of 𝑋(𝜔) peals at 𝜔 = 0 and decays at higher
𝑡
frequencies. Therefore, 𝑟𝑒𝑐𝑡 is a lowpass signal with most of the
𝜏
signal energy in lower-frequency components.
• Strictly speaking, because the spectrum extends from 0 to ∞. The
bandwidth is ∞.
• However, must of the spectrum is concentrated within first lobe (from
ω = 0 to ω = 2π/τ). Therefore, a rough estimate of the bandwidth of a
rectangular pulse of width τ seconds is 2π/τ rad/s, or 1/tau Hz[†].
• Note the reciprocal relationship of the pulse width with its
bandwidth.

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 25

 Example 7.3
Fourier Transform of Dirac Delta Function 𝛿(𝑡)
∞
ℱ𝛿 𝑡 = න 𝛿 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = 1 Therefore, 𝛿 𝑡 ⟺ 1
−∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 26

Example 7.4
Inverse Fourier Transform of the Dirac Delta Function 𝛿(𝜔)

1 ∞ 1
ℱ −1 𝛿 𝜔 = 𝑗𝜔𝑡
න 𝛿 𝜔 𝑒 𝑑𝜔 =
2𝜋 −∞ 2𝜋

Therefore,
1
⟺𝛿 𝜔 and 1 ⟺ 2𝜋𝛿(𝜔)
2𝜋

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 27

Example 7.4
Inverse Fourier Transform of the Dirac Delta Function 𝛿(𝜔)
Frequency of a DC signal

If this was shifted,

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 28

Example 7.5
The inverse Fourier transform of shifted Dirac Delta δ(ω − ω0)

1 ∞ 1 𝑗𝜔 𝑡
ℱ −1 𝛿 𝜔−𝜔0 = 𝑗𝜔𝑡
න 𝛿 𝜔−𝜔0 𝑒 𝑑𝜔 = 𝑒 0
2𝜋 −∞ 2𝜋

Therefore, shifted delta function

1 𝑗𝜔 𝑡
𝑒 0 ⟺ 𝛿 𝜔 − 𝜔0 and 𝑒 𝑗𝜔0 𝑡 ⟺ 2𝜋𝛿(𝜔 − 𝜔0 )
2𝜋

It also follows that,

1 −𝑗𝜔 𝑡
𝑒 0 ⟺ 𝛿 𝜔 +𝜔
0 and 𝑒 −𝑗𝜔0 𝑡 ⟺ 2𝜋𝛿(𝜔 + 𝜔0 )
2𝜋
MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 29
 Example 7.6
Fourier Transform of a Sinusoid cos 𝜔0 𝑡
• From Euler’s formula
1
cos 𝜔0 𝑡 = (𝑒 𝑗𝜔0 𝑡 + 𝑒 −𝑗𝜔0 𝑡 )
2

cos 𝜔0 𝑡 ⟺ 𝜋[𝛿 𝜔 + 𝜔0 + 𝛿(𝜔 − 𝜔0 )]

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 30

 Example 7.7
Fourier Transform of a Periodic Signal
• The Fourier series of a periodic signal 𝑥(𝑡) with period 𝑇0 is given by
∞

𝑥 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0 𝑡
𝑛=−∞

• Taking the Fourier transform of both sides, we obtain

∞

𝑋 𝜔 = 2𝜋 ෍ 𝐷𝑛 𝛿(𝜔 − 𝜔0 )
𝑛=−∞

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 31

 Example 7.8
Fourier Transform of a Dirac Delta Train

MENG 3520 Signal Processing Instructor: Dr. Nilufar Somayyeh Poshtiban 32