Tex as A& M University

ECEN 314– Spring 2022

Signals and Systems

Module 12

Fourier Transforms
Learning Objectives

Be able to find Fourier transforms of signals that contain

delta functions in either the time domain or the frequency
domain.
Understand the difference between a power and an energy
signal.
Be able to compute the power/energy of a signal directly or
using Parseval’s theorem.
Understand the various definitions of bandwidth and be able
to compute the bandwidth of typical signals.
Fourier Transform of Delta
Functions
When taking Fourier transforms involving delta functions, it is helpful to use the
sifting property of delta functions:
𝑏𝑏
𝑥𝑥 𝑡𝑡𝑜𝑜 , 𝑎𝑎 < 𝑡𝑡𝑜𝑜 < 𝑏𝑏,
� 𝛿𝛿 𝑡𝑡 − 𝑡𝑡𝑜𝑜 𝑥𝑥 𝑡𝑡 𝑑𝑑𝑑𝑑 = �
𝑎𝑎 0, 𝑡𝑡𝑜𝑜 < 𝑎𝑎, or 𝑡𝑡𝑜𝑜 > 𝑏𝑏.

Consider taking the Fourier Transform of the delta function

∞
ℱ 𝛿𝛿 𝑡𝑡 = ∫−∞ 𝛿𝛿 𝑡𝑡 𝑒𝑒 −𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗 𝑑𝑑𝑑𝑑 = 1.

Likewise, with the inverse Fourier Transform,

∞
ℱ −1 𝛿𝛿 𝑓𝑓 = ∫−∞ 𝛿𝛿 𝑓𝑓 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗 𝑑𝑑𝑑𝑑 = 1.

Therefore, if 𝑥𝑥 𝑡𝑡 = 𝛿𝛿(𝑡𝑡) is input to a circuit/system with transfer function 𝐻𝐻(𝑓𝑓),

then the output will be 𝑌𝑌 𝑓𝑓 = 𝐻𝐻 𝑓𝑓 𝑋𝑋 𝑓𝑓 = 𝐻𝐻 𝑓𝑓 (since 𝑋𝑋 𝑓𝑓 = 1 in this case) so
that 𝑦𝑦 𝑡𝑡 = ℱ −1 𝐻𝐻 𝑓𝑓 = ℎ(𝑡𝑡). This is why ℎ(𝑡𝑡) is called the impulse response as it
is the output that appears when the input is a unit impulse function.
Fourier Transforms of Sinusoids
Starting with ℱ −1 𝛿𝛿 𝑓𝑓 = 1, and then using the frequency shifting property:
ℱ −1 𝛿𝛿 𝑓𝑓 − 𝑓𝑓𝑜𝑜 = 𝑒𝑒 −𝑗𝑗𝑗𝑗𝑗𝑓𝑓𝑜𝑜 𝑡𝑡 ,
ℱ −1 𝛿𝛿 𝑓𝑓 + 𝑓𝑓𝑜𝑜 = 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑓𝑓𝑜𝑜 𝑡𝑡 ,
and then the linearity property,
1 1 1 1
ℱ −1 𝛿𝛿 𝑓𝑓 + 𝑓𝑓𝑜𝑜 + 𝛿𝛿 𝑓𝑓 − 𝑓𝑓𝑜𝑜 = 𝑒𝑒 𝑗𝑗𝑗𝑗𝑗𝑓𝑓𝑜𝑜 𝑡𝑡 + 𝑒𝑒 −𝑗𝑗𝑗𝑗𝑗𝑓𝑓𝑜𝑜 𝑡𝑡 = cos 2𝜋𝜋𝑓𝑓𝑜𝑜 𝑡𝑡 ,
2 2 2 2
1 1 1 𝑗𝑗𝑗𝑗𝑗𝑓𝑓 𝑡𝑡 1 −𝑗𝑗𝑗𝑗𝑗𝑓𝑓 𝑡𝑡
ℱ −1 𝛿𝛿 𝑓𝑓 + 𝑓𝑓𝑜𝑜 − 𝛿𝛿 𝑓𝑓 − 𝑓𝑓𝑜𝑜 = 𝑒𝑒 𝑜𝑜 − 𝑒𝑒 𝑜𝑜 = sin 2𝜋𝜋𝑓𝑓 𝑡𝑡 .
𝑜𝑜
2𝑗𝑗 2𝑗𝑗 2𝑗𝑗 2𝑗𝑗
Therefore,
1 1
ℱ cos 2𝜋𝜋𝑓𝑓𝑜𝑜 𝑡𝑡 = 𝛿𝛿 𝑓𝑓 + 𝑓𝑓𝑜𝑜 + 𝛿𝛿 𝑓𝑓 − 𝑓𝑓𝑜𝑜 ,
2 2

𝑗𝑗 𝑗𝑗
ℱ sin 2𝜋𝜋𝑓𝑓𝑜𝑜 𝑡𝑡 = 𝛿𝛿 𝑓𝑓 − 𝑓𝑓𝑜𝑜 − 𝛿𝛿 𝑓𝑓 + 𝑓𝑓𝑜𝑜 .
2 2
Not surprisingly, all of the frequency content of sinusoids is contained at 𝑓𝑓 = ±𝑓𝑓𝑜𝑜 .
Fourier Transforms of Periodic
Signals
Suppose 𝑥𝑥(𝑡𝑡) is a periodic signal with a period of 𝑇𝑇𝑜𝑜 (fundamental frequency of 𝑓𝑓𝑜𝑜 =
1/𝑇𝑇𝑜𝑜 ) with some Fourier series representation
∞

𝑥𝑥 𝑡𝑡 = � 𝑋𝑋𝑘𝑘 𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝑓𝑓𝑜𝑜 𝑡𝑡 .
𝑘𝑘=−∞

Taking F.T.’s of both sides of the equation,

∞ ∞ ∞

ℱ 𝑥𝑥 𝑡𝑡 =ℱ � 𝑋𝑋𝑘𝑘 𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝑘𝑘𝑓𝑓𝑜𝑜 𝑡𝑡 = � 𝑋𝑋𝑘𝑘 ℱ 𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝑘𝑘𝑓𝑓𝑜𝑜 𝑡𝑡 = � 𝑋𝑋𝑘𝑘 𝛿𝛿 𝑓𝑓 − 𝑘𝑘𝑓𝑓𝑜𝑜 .

𝑘𝑘=−∞ 𝑘𝑘=−∞ 𝑘𝑘=−∞

Periodic signals have a Fourier transform consisting of discrete components (not a

continuum of frequency components). We refer to this as a line spectrum.

𝑋𝑋(𝑓𝑓)
𝑥𝑥(𝑡𝑡) 𝑋𝑋−1 𝑋𝑋1
𝑋𝑋−3 𝑋𝑋3
... ... 𝑋𝑋−5 𝑋𝑋−2 𝑋𝑋2 𝑋𝑋5
𝑓𝑓
−𝑇𝑇𝑜𝑜 𝑇𝑇𝑜𝑜 𝑡𝑡 −5𝑓𝑓𝑜𝑜 −3𝑓𝑓𝑜𝑜 −𝑓𝑓𝑜𝑜 𝑓𝑓𝑜𝑜 3𝑓𝑓𝑜𝑜 5𝑓𝑓𝑜𝑜
𝑋𝑋−4 𝑋𝑋4
The Impulse Train
Later in the course we will use a periodic 𝑥𝑥(𝑡𝑡)
impulse train to convert signals from
continuous time to discrete time.
∞
... ...
𝑥𝑥 𝑡𝑡 = � 𝛿𝛿(𝑡𝑡 − 𝑚𝑚𝑡𝑡𝑜𝑜 ) .
𝑚𝑚=−∞ −2𝑇𝑇𝑜𝑜 −𝑇𝑇𝑜𝑜 𝑇𝑇𝑜𝑜 2𝑇𝑇𝑜𝑜 3𝑇𝑇𝑜𝑜 𝑡𝑡
Since this signal is periodic, it has a Fourier
Series representation with coefficients given
by
𝑋𝑋(𝑓𝑓)
𝑇𝑇𝑜𝑜 /2
1
𝑋𝑋𝑘𝑘 = � 𝑥𝑥 𝑡𝑡 𝑒𝑒 −𝑗𝑗𝑗𝜋𝜋𝑘𝑘𝑘𝑘𝑜𝑜 𝑡𝑡 𝑑𝑑𝑑𝑑
𝑇𝑇𝑜𝑜 −𝑇𝑇𝑜𝑜/2 1/𝑇𝑇𝑜𝑜
𝑇𝑇𝑜𝑜 /2
... ...
1 1
= � 𝛿𝛿 𝑡𝑡 𝑒𝑒 −𝑗𝑗𝑗𝜋𝜋𝑘𝑘𝑘𝑘𝑜𝑜𝑡𝑡 𝑑𝑑𝑑𝑑 = . 𝑡𝑡
𝑇𝑇𝑜𝑜 −𝑇𝑇𝑜𝑜/2 𝑇𝑇𝑜𝑜 −2𝑓𝑓𝑜𝑜 −𝑓𝑓𝑜𝑜 𝑓𝑓𝑜𝑜 2𝑓𝑓𝑜𝑜 3𝑓𝑓𝑜𝑜

Therefore, the F.T. of the impulse train is

itself an impulse train.
Fourier Transform of Unit Step
𝑠𝑠𝑠𝑠𝑠𝑠(𝑡𝑡)
Start with the signum function, 1
1, 𝑡𝑡 > 0,
𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡 = �
−1, 𝑡𝑡 < 0. 𝑡𝑡
𝑑𝑑 -1
It’s derivative is 𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡 = 2𝛿𝛿(𝑡𝑡).
𝑑𝑑𝑑𝑑
Therefore,
𝑑𝑑
𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡 ↔ 2. 𝑢𝑢(𝑡𝑡)
𝑑𝑑𝑑𝑑 1
Using the integration property of F.T.’s, then

𝑠𝑠𝑠𝑠𝑠𝑠 𝑡𝑡 ↔
2
(or
1
). 𝑡𝑡
𝑗𝑗𝑗𝑗 𝑗𝑗𝜋𝜋𝜋𝜋 -1
We can relate the unit step function to the
signum function according to
Note: The presence of the delta
1 𝑠𝑠𝑠𝑠𝑠𝑠(𝑡𝑡)
𝑢𝑢 𝑡𝑡 = + function in the spectrum of the unit
2 2 step is the result of the step having a
1 1 1 1
ℱ 𝑢𝑢 𝑡𝑡 = 2 𝛿𝛿(𝜔𝜔) + 𝑗𝑗𝑗𝑗 or 𝛿𝛿 𝑓𝑓 + 𝑗𝑗2𝜋𝜋𝜋𝜋 . DC component (average value) of ½.
2
 Duality

o Another very useful property of Fourier transforms is called duality and

involves the exchange of the roles of time and frequency in the Fourier
transform relationship.
∞
o Start with FT relationship: 𝑥𝑥 𝑡𝑡 = ∫−∞ 𝑋𝑋(𝑓𝑓)𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝜋𝜋 𝑑𝑑𝑑𝑑  𝑥𝑥(𝑡𝑡) ↔ 𝑋𝑋(𝑓𝑓)
∞
o Negate time: 𝑥𝑥 −𝑡𝑡 = ∫−∞ 𝑋𝑋(𝑓𝑓)𝑒𝑒 −𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝜋𝜋 𝑑𝑑𝑑𝑑

o Exchange roles of the time and frequency variables, 𝑡𝑡 → 𝑓𝑓, 𝑓𝑓 → 𝑡𝑡:

∞
𝑥𝑥 −𝑓𝑓 = ∫−∞ 𝑋𝑋(𝑡𝑡)𝑒𝑒 −𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝜋𝜋 𝑑𝑑𝑡𝑡  𝑋𝑋(𝑡𝑡) ↔ 𝑥𝑥(−𝑓𝑓)

o In words, this means that if we know a certain Fourier transform pair, then we
can exchange the roles of time and frequency to produce a new Fourier
transform pair (to within a minus sign).
Example
rect(𝑡𝑡) ↔ sinc(𝑓𝑓) duality sinc(𝑡𝑡) ↔ rect(−𝑓𝑓)
 Power/Energy in Signals

o Recall from your study of circuits, that if 𝑥𝑥(𝑡𝑡) is a voltage in a resistor, then its
instantaneous power is given by 𝑝𝑝 𝑡𝑡 = 𝑥𝑥 2 (𝑡𝑡)/𝑅𝑅 while if it is a current then
𝑝𝑝 𝑡𝑡 = 𝑥𝑥 2 𝑡𝑡 ∗ 𝑅𝑅.
o In order to allow us to study the power in a signal without having to worry
about whether it is a current or a voltage, we often talk about a normalized
power where it is assumed that 𝑅𝑅 = 1Ω, so that regardless of whether the
signal is a voltage or a current, 𝑝𝑝 𝑡𝑡 = 𝑥𝑥 2 (𝑡𝑡).
𝑑𝑑𝑑𝑑
o If we use the relationship between power and energy, 𝑝𝑝 𝑡𝑡 = , then energy
𝑑𝑑𝑑𝑑
in this context energy would be 𝑤𝑤 𝑡𝑡 = ∫ 𝑝𝑝 𝑡𝑡 𝑑𝑑𝑑𝑑 = ∫ 𝑥𝑥 2 𝑡𝑡 𝑑𝑑𝑑𝑑.

o To make these expressions not functions of time, we consider

1 𝑇𝑇𝑜𝑜 2
 Average Power = 𝑃𝑃 = ∫ 𝑥𝑥 𝑡𝑡 𝑑𝑑𝑑𝑑 (for periodic signals)
𝑇𝑇𝑜𝑜 𝑡𝑡𝑜𝑜
∞
 Total Energy = 𝑊𝑊 = ∫−∞ 𝑥𝑥 2 𝑡𝑡 𝑑𝑑𝑑𝑑 (for aperiodic finite duration signals)

o Moving forward, we will just refer to these quantities as power and energy.
 Power/Energy in Signals

o Generally, a signal that is periodic will have 𝑃𝑃 =finite but 𝑊𝑊 = ∞. We refer to

this as a power type signal as it is meaningful to talk about its power but not
its energy.
o Generally, a signal that is aperiodic will have 𝑃𝑃 =0 but 𝑊𝑊 =finite. We refer to
this as an energy type signal as it is meaningful to talk about its energy but
not its power.

2 2
... ...
1 1
−1 1 3 5 𝑡𝑡 −1 1 3 5 𝑡𝑡
−1 −1

A power type signal An energy type signal

𝑃𝑃 = 1.5, 𝑊𝑊 = ∞. 𝑃𝑃 = 0, 𝑊𝑊 = 6.
 Parseval’s Theorem
o It turns out that we can determine the energy (or power) of a signal from its
frequency content.
∞ ∞ ∞
𝑊𝑊 = � 𝑥𝑥 2 (𝑡𝑡) 𝑑𝑑𝑑𝑑 = � 𝑥𝑥(𝑡𝑡) � 𝑋𝑋(𝑓𝑓) 𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝜋𝜋 𝑑𝑑𝑑𝑑 𝑑𝑑𝑡𝑡
−∞ −∞ −∞
∞ ∞ ∞ ∞
= � 𝑋𝑋(𝑓𝑓) � 𝑥𝑥(𝑡𝑡) 𝑒𝑒 𝑗𝑗𝑗𝜋𝜋𝜋𝜋𝜋𝜋 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 = � 𝑋𝑋 𝑓𝑓 𝑋𝑋 ∗ 𝑓𝑓 𝑑𝑑𝑑𝑑 = � 𝑋𝑋 𝑓𝑓 2 𝑑𝑑𝑑𝑑
−∞ −∞ −∞ −∞

o This result is known as Parseval’s theorem and it states that the energy of a
signal is the same in the time domain and in the frequency domain
∞ ∞ 1 ∞
𝑊𝑊 = ∫−∞ 𝑥𝑥 2 (𝑡𝑡) = ∫−∞ 𝑋𝑋 𝑓𝑓 2 𝑑𝑑𝑑𝑑
or ∫ 𝑋𝑋 𝜔𝜔 2 𝑑𝑑𝜔𝜔
2𝜋𝜋 −∞

o For power type signals, there is an equivalent result involving Fourier series
coefficients
𝑇𝑇𝑜𝑜 ∞
1
𝑃𝑃 = � 𝑥𝑥 2 𝑡𝑡 𝑑𝑑𝑑𝑑 = � 𝑋𝑋𝑘𝑘 2
𝑇𝑇𝑜𝑜
𝑡𝑡𝑜𝑜 𝑘𝑘=−∞
 Example
o Suppose we wanted to find the energy in the signal 𝑥𝑥 𝑡𝑡 = sinc(𝑡𝑡).
Performing the computation directly, we would need to calculate the integral
∞ ∞ sin2 (𝜋𝜋𝜋𝜋)
𝑊𝑊 = ∫−∞ sinc 2 (𝑡𝑡) 𝑑𝑑𝑑𝑑 = ∫−∞ (𝜋𝜋𝜋𝜋)2 𝑑𝑑𝑑𝑑.

o This integral would be rather tricky to do.

o Instead we can use Parseval’s theorem together with the fact that sinc(𝑡𝑡) ↔
rect(𝑓𝑓) so that
∞ ∞
𝑊𝑊 = � sinc 2 (𝑡𝑡) 𝑑𝑑𝑑𝑑 = � rect 2 (𝑓𝑓) 𝑑𝑑𝑓𝑓 = 1
−∞ −∞
1 1

0.8 0.8

0.6
Areas underneath 0.6
(t)

(f)
2

2
two curves are equal
rect
sinc

0.4 0.4

0.2 0.2

0 0

-5 0 5 -1 -0.5 0 0.5 1

time frequency
 Time/Frequency Relationships
𝑥𝑥(𝑡𝑡) 𝑋𝑋(𝑓𝑓)
Consider the time-scaling
property of Fourier Transforms:
1 𝑓𝑓 𝑡𝑡 𝑓𝑓
𝑥𝑥 𝑎𝑎𝑎𝑎 ↔ 𝑋𝑋 .
𝑎𝑎 𝑎𝑎
Suppose 𝑥𝑥(𝑡𝑡) has a bandwidth 𝑥𝑥(2𝑡𝑡) 𝑋𝑋(𝑓𝑓/2)/2
of 𝐵𝐵 Hz so that 𝑋𝑋 𝑓𝑓 = 0 for all
𝑓𝑓 > 𝐵𝐵.
𝑓𝑓 𝑓𝑓
Then, 𝑋𝑋 = 0 for all > 𝐵𝐵 𝑡𝑡 𝑓𝑓
𝑎𝑎 𝑎𝑎
or equivalently, for all 𝑓𝑓 > 𝑎𝑎𝐵𝐵.
Therefore, 𝑥𝑥 𝑎𝑎𝑎𝑎 will have a
bandwidth of 𝑎𝑎𝑎𝑎 Hz. 2𝑋𝑋(2𝑓𝑓)

 If 𝑎𝑎 > 1, compression in time 𝑥𝑥(𝑡𝑡/2)

↔ expansion in frequency,
 If 𝑎𝑎 < 1, compression in time 𝑓𝑓
↔ expansion in frequency. 𝑡𝑡
 Time and Bandlimited Signals
o Suppose a signal is time-limited so that 𝑥𝑥(𝑡𝑡) ≠ 0 only for some range 𝑡𝑡1 < 𝑡𝑡 < 𝑡𝑡2 .
Furthermore, suppose the signal is bandlimited to some range −𝐵𝐵 < 𝑓𝑓 < 𝐵𝐵.
o Next suppose we multiply 𝑥𝑥(𝑡𝑡) by a rectangle that is equal to 1 only over 𝑡𝑡1 < 𝑡𝑡 < 𝑡𝑡2 . Then,
by the multiplication property of F.T.s
𝑡𝑡1 + 𝑡𝑡2
𝑡𝑡 − 𝑡𝑡 +𝑡𝑡
𝑥𝑥 𝑡𝑡 � rect 2 ↔ 𝑋𝑋 𝑓𝑓 ∗ 𝑡𝑡2 − 𝑡𝑡1 sinc 𝑡𝑡2 − 𝑡𝑡1 𝑓𝑓 𝑒𝑒
−𝑗𝑗𝑗𝑗𝑗𝑗𝑗 1 2
2
𝑡𝑡2 − 𝑡𝑡1
o In the time domain, the signal is unchanged since we are multiplying by 1 when the signal is
non-zero and 0 when the signal is zero.
o In the frequency domain, convolving something with a frequency duration of 2𝐵𝐵 with
something that has an infinite frequency duration will produce a result that has an infinite
frequency duration.
o The only way this can be logically consistent is if 𝐵𝐵 = ∞.

 No signal can be simultaneously time-limited and frequency-limited. If it is time-

limited it must have infinite frequency duration, if it is frequency-limited, it must have
infinite time duration.
 Since we are mortal beings, we cannot create signals of infinite time duration.
 Strictly speaking, any practical signal we create must be of infinite frequency duration.
 Bandwidth
o The bandwidth of a signal measures how wide a signal is in the frequency
domain. There are several possible ways to measure bandwidth;
 Absolute BW: 𝐵𝐵𝑎𝑎𝑎𝑎𝑎𝑎 is the absolute BW of 𝑥𝑥(𝑡𝑡) if 𝐵𝐵𝑎𝑎𝑎𝑎𝑎𝑎 is the smallest
number such that 𝑋𝑋 𝑓𝑓 = 0 for all 𝑓𝑓 > 𝐵𝐵𝑎𝑎𝑎𝑎𝑎𝑎 .
 Null BW: 𝐵𝐵𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 is the null BW of 𝑥𝑥(𝑡𝑡) if 𝐵𝐵𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 is the smallest number such
that 𝑋𝑋 𝐵𝐵𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 = 0.
 P% Energy BW: This is the width of the frequency band which contains
P% of the total signal energy.
 Equivalent BW: This will be the bandwidth of a signal with a flat spectrum
and same peak spectrum with the same amount of energy as the signal in
question. Mathematically it is
∞ 𝑋𝑋(𝑓𝑓) 2
𝐵𝐵𝑒𝑒𝑒𝑒 = ∫0 𝑋𝑋 𝑑𝑑𝑑𝑑.
𝑚𝑚𝑚𝑚𝑚𝑚

 P% Level BW: 𝐵𝐵𝑙𝑙𝑙𝑙𝑙𝑙 is the P% Level BW of 𝑥𝑥(𝑡𝑡) if 𝐵𝐵𝑙𝑙𝑙𝑙𝑙𝑙 is the smallest

𝑃𝑃
number such that 𝑋𝑋 𝑓𝑓 < 𝑋𝑋𝑚𝑚𝑚𝑚𝑚𝑚 for all 𝑓𝑓 > 𝐵𝐵𝑙𝑙𝑙𝑙𝑙𝑙 .
100
 Bandwidth
Example: 𝑥𝑥 𝑡𝑡 = 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑡𝑡 ↔ 𝑋𝑋 𝑓𝑓 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠(𝑓𝑓), |𝑋𝑋 𝑓𝑓 |2 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 2 (𝑓𝑓), 𝑋𝑋𝑚𝑚𝑚𝑚𝑚𝑚 = 1.
1

𝐵𝐵𝑎𝑎𝑎𝑎𝑎𝑎 = ∞

0.8

𝐵𝐵𝑒𝑒𝑒𝑒 = 0.5 𝐻𝐻𝐻𝐻

𝐵𝐵𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒(90%)
0.6
= 0.868𝐻𝐻𝐻𝐻
|X(f)|

𝐵𝐵𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒(95%)
0.4
= 2.20𝐻𝐻𝐻𝐻
𝐵𝐵𝑙𝑙𝑙𝑙𝑙𝑙(10%) = 2.68 𝐻𝐻𝐻𝐻
𝐵𝐵𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒(99%)
0.2
= 11.4𝐻𝐻𝐻𝐻
10% level
0

-5 -4 -3 -2 -1 0 1 2 3 4 5

f (Hz)
Note: When measuring
BWs, only measure
𝐵𝐵𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 = 1 𝐻𝐻𝐻𝐻 positive frequencies.