ELEC1200: A System View of

Communications: from Signals to Packets

Lecture 14
• Motivation
– Music as a combination of sounds at different frequencies

• Continuous-time sinusoids

• Discrete-time sinusoids

• Decomposing signals into a sum of cosines

– Amplitude Spectrum
ELEC1200
 Motivation: Music
Music is a combination of sounds with different frequencies.
In this lecture, we see that this is true of any waveform
combination
1

-1
0 5 10 15 20 25 30
time in milliseconds

frequency = 261.63Hz (C4) frequency = 329.63Hz (E4) frequency = 392.00Hz (G4)

1 1 1

0 0 0

-1 -1 -1
0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
time in milliseconds time in milliseconds time in milliseconds

ELEC1200 increasing frequency 2

 Visualizing the frequency content of music
amplitude

frequency
ELEC1200 https://www.youtube.com/watch?v=iAi3vi2j7N0 3
 ELEC1200: A System View of
Communications: from Signals to Packets
Lecture 14
• Motivation
– Music as a combination of sounds at different frequencies

• Continuous-time sinusoids

• Discrete-time sinusoids

• Decomposing signals into a sum of cosines

– Amplitude Spectrum

ELEC1200 4
 Sinusoidal Waves

Cosine Wave Sine Wave

2 2
𝑥𝑥(𝑡𝑡) = cos( 2𝜋𝜋𝑓𝑓𝑡𝑡) 𝑥𝑥(𝑡𝑡) = sin( 2𝜋𝜋𝑓𝑓𝑡𝑡)
1 1

0 0

-1 -1

-2 -2
𝑇𝑇/2 𝑇𝑇 3𝑇𝑇/2 2𝑇𝑇 𝑇𝑇/2 𝑇𝑇 3𝑇𝑇/2 2𝑇𝑇
t (sec) t (sec)

Parameters:
𝒇𝒇 = Frequency in hertz (cycles per second)
𝑻𝑻 = Period (in seconds) = 𝒇𝒇−𝟏𝟏

ELEC1200 5
 Phase
cos 2𝜋𝜋𝑓𝑓𝑡𝑡 + 𝜙𝜙
• It is useful to add a phase parameter, 1
𝜙𝜙 = −𝜋𝜋/2
𝜙𝜙 ∈ 0,2𝜋𝜋 or −𝜋𝜋, 𝜋𝜋 , 0
-1
which shifts the cosine wave back and forth in 0 𝑇𝑇�
4
𝑇𝑇�
2
3𝑇𝑇�
4 𝑇𝑇
time. 1
𝜙𝜙 = −𝜋𝜋/4
0
-1
• Recall that cos(𝑡𝑡) reaches its maximum value at 0 𝑡𝑡0 𝑇𝑇� 𝑇𝑇� 3𝑇𝑇� 𝑇𝑇
4 2 4
0, i.e., cos 0 = 1 . 1
0 𝜙𝜙 = 0
• Thus, cos( 2𝜋𝜋𝑓𝑓𝑡𝑡 + 𝜙𝜙) reaches its maximum value -1
0 𝑇𝑇� 𝑇𝑇� 3𝑇𝑇� 𝑇𝑇
4 2 4
at 𝑡𝑡0 , where 1
2𝜋𝜋𝜋𝜋𝑡𝑡0 + 𝜙𝜙 = 0 0
𝜙𝜙 = 𝜋𝜋/4
-1
0 𝑇𝑇� 𝑇𝑇� 3𝑇𝑇�
𝜙𝜙 4 2 4 𝑇𝑇
𝑡𝑡0 = − 1
2𝜋𝜋𝜋𝜋
0
𝜙𝜙 = 𝜋𝜋/2
-1
0 𝑇𝑇�
4
𝑇𝑇�
2
3𝑇𝑇�
4 𝑇𝑇

t (sec)

ELEC1200 6
 Sine is a shifted cosine
cos 2𝜋𝜋𝑓𝑓𝑡𝑡 + 𝜙𝜙
• Recall that 1
𝜙𝜙 = −𝜋𝜋/2
cos(𝐴𝐴 + 𝐵𝐵) = cos(𝐴𝐴)cos(𝐵𝐵) – sin(𝐴𝐴)sin(𝐵𝐵). 0
-1
• Thus, 0 𝑇𝑇�
4
𝑇𝑇�
2
3𝑇𝑇�
4 𝑇𝑇
= cos 2𝜋𝜋𝜋𝜋𝑡𝑡 cos 𝜙𝜙 − sin 2𝜋𝜋𝑓𝑓𝑡𝑡 sin 𝜙𝜙 1
𝜙𝜙 = −𝜋𝜋/4
0
-1
• Since cos −𝜋𝜋2 = 0 and sin −𝝅𝝅𝟐𝟐 = −1, 0 𝑡𝑡0 𝑇𝑇�
4
𝑇𝑇�
2
3𝑇𝑇�
4 𝑇𝑇

1
𝜋𝜋
cos 2π𝑓𝑓t − = cos 2π𝑓𝑓𝑡𝑡 � 0 − sin 2π𝑓𝑓𝑡𝑡 � −1 0 𝜙𝜙 = 0
2
-1
0 𝑇𝑇� 𝑇𝑇� 3𝑇𝑇� 𝑇𝑇
= sin 2π𝑓𝑓𝑡𝑡 4 2 4
1
• Thus, a sine wave is just a phase-shifted cosine. 0
𝜙𝜙 = 𝜋𝜋/4
-1
0 𝑇𝑇� 𝑇𝑇� 3𝑇𝑇� 𝑇𝑇
4 2 4
1
• Only need to work with cosines (with phase shift)! 0
𝜙𝜙 = 𝜋𝜋/2
-1
0 𝑇𝑇�
4
𝑇𝑇�
2
3𝑇𝑇�
4 𝑇𝑇

t (sec)

ELEC1200 7
 General Form
• Any sinusoidal signal can be expressed as 𝐴𝐴 cos( 2𝜋𝜋𝑓𝑓𝑡𝑡 + 𝜙𝜙)

𝐴𝐴 cos( 2𝜋𝜋𝑓𝑓𝑡𝑡 + 𝜙𝜙)

2

1 𝐴𝐴 = amplitude
𝐴𝐴 𝑓𝑓 = frequency
𝜙𝜙 = phase
0 𝑇𝑇 = period
𝑡𝑡0
𝜙𝜙
𝑡𝑡0 = −
-1
2𝜋𝜋𝜋𝜋
𝑇𝑇 = 𝑓𝑓 −1

-2
-1 -0.5 0 0.5 1

ELEC1200 8
 ELEC1200: A System View of
Communications: from Signals to Packets
Lecture 14
• Motivation
– Music as a combination of sounds at different frequencies

• Continuous-time sinusoids

• Discrete-time sinusoids

• Decomposing signals into a sum of cosines

– Amplitude Spectrum

ELEC1200 9
 Continuous and Discrete Signals

1. Signal: x(t)=A cos(ωt+φ)

15

10

2. Continuous time (Analog)

5

0
0 5 10 15 20
-5

-10

-15

x[n]

3. Discrete time (Digital) n

ELEC1200 10
 Discrete Time Cosines
• In this course, we consider only discrete time cosine waveforms with a finite
number of samples, N (typically a power of 2), indexed by 𝒏𝒏 ∈ 𝟎𝟎, 𝟏𝟏, 𝟐𝟐, … , 𝑵𝑵 − 𝟏𝟏 .
• A discrete time cosine has the form
𝑘𝑘 𝑁𝑁
𝑥𝑥 𝑛𝑛 = cos 2𝜋𝜋 𝑛𝑛 + 𝜙𝜙 where 𝑘𝑘 ∈ 0,1, … , and 𝜙𝜙 ∈ [0,2𝜋𝜋) or [−𝜋𝜋, 𝜋𝜋)
𝑁𝑁 2

𝑁𝑁 = 32
1

0.5
𝑘𝑘 = 4
0 𝜙𝜙 = 0
-0.5

-1

0 5 10 15 20 25 30 n

• The variable 𝑘𝑘 is a discrete time frequency, which has units of cycles/waveform.

• Since the waveform has N samples, the period (in samples) is

𝑁𝑁 samples waveform 𝑵𝑵 samples

period = ⋅ =
waveform 𝑘𝑘 cycles 𝒌𝒌 cycle
ELEC1200 11
 Discrete time cosines with varying k
k=1
𝑘𝑘 1
period = 64
𝑥𝑥 𝑛𝑛 = cos 2𝜋𝜋 𝑛𝑛 + 𝜙𝜙
𝑁𝑁 0

-1
n

𝑁𝑁 = 64 0 10 20 30 40 50 60

k=2
1

𝜙𝜙 = 0
𝑁𝑁
0
period = 32
period =
𝑘𝑘
-1
n
0 10 20 30 40 50 60

k=4
0

-1
n period = 16
0 10 20 30 40 50 60

-1
n k=8
period= 8
0 10 20 30 40 50 60

ELEC1200 12
 Relationship between Discrete and Continuous Time
𝑘𝑘
• Suppose that 𝑥𝑥 𝑛𝑛 = cos 2𝜋𝜋 𝑛𝑛 is obtained by sampling a continuous time cosine at
𝑁𝑁
1
a sampling frequency of 𝐹𝐹𝑠𝑠 (sample period 𝑇𝑇𝑠𝑠 = ).
𝐹𝐹𝑠𝑠
𝑡𝑡
• Since the 𝑛𝑛th sample is at time 𝑡𝑡 = 𝑛𝑛 ⋅ 𝑇𝑇𝑠𝑠 (𝑛𝑛 = ). Thus, the continuous time
𝑇𝑇𝑠𝑠
1
waveform is 𝑇𝑇𝑠𝑠 =
𝐹𝐹𝑠𝑠

𝑘𝑘 𝑘𝑘 𝑡𝑡 𝑘𝑘 𝑘𝑘
𝑥𝑥𝑐𝑐 𝑡𝑡 = cos 2𝜋𝜋 𝑛𝑛 � = cos 2𝜋𝜋 = cos 2𝜋𝜋 𝐹𝐹 𝑡𝑡 = cos 2𝜋𝜋𝑓𝑓𝑡𝑡 where 𝑓𝑓 = 𝐹𝐹
𝑁𝑁 𝑡𝑡
𝑛𝑛=𝑇𝑇 𝑁𝑁 𝑇𝑇𝑠𝑠 𝑁𝑁 𝑠𝑠 𝑁𝑁 𝑠𝑠
𝑠𝑠

𝑁𝑁 = 32
1

• If 𝐹𝐹𝑠𝑠 = 64kHz, 0.5

𝑘𝑘 = 4
𝑘𝑘
𝑓𝑓 = 𝐹𝐹 𝜙𝜙 = 0
𝑁𝑁 𝑠𝑠
0

4
= 64 kHz
-0.5

32
= 8 kHz
-1

0 5 10 15 20 25 30
n
1
𝑇𝑇𝑠𝑠 = = 16μs
𝐹𝐹𝑠𝑠 32 32
32 samples = 32 ⋅ 𝑇𝑇𝑠𝑠 = = = 0.5ms
𝐹𝐹𝑠𝑠 64000
ELEC1200 13
 ELEC1200: A System View of
Communications: from Signals to Packets
Lecture 14
• Motivation
– Music as a combination of sounds at different frequencies

• Continuous-time sinusoids

• Discrete-time sinusoids

• Decomposing signals into a sum of cosines

– Amplitude Spectrum

ELEC1200 14
 Sinusoidal Representation of Sampled Data
• Any sampled data waveform x(n) with N samples:
x(n)
= for n 0,1,  (N − 1)

can be expressed as the sum of a N/2+1 cosine waves using the equation

𝑁𝑁/2 each term in the

𝑘𝑘 summation is called a
𝑥𝑥(𝑛𝑛) = � 𝐴𝐴𝑘𝑘 cos( 2𝜋𝜋 𝑛𝑛 + 𝜙𝜙𝑘𝑘 ) frequency component (tone)
𝑁𝑁
𝑘𝑘=0

• This equation is called the Fourier series expansion of x(n)

• Different waveforms are obtained by changing the values of Ak and φk
– Since for k=0, the term in the summation is 𝐴𝐴0 cos( 𝜙𝜙0 ), we usually assume 𝜙𝜙0 = 0.

ELEC1200 15
 Interpretation
1 20
𝑁𝑁/2 1 + 0.8 cos( 2𝜋𝜋 𝑛𝑛) + 0.1 cos( 2𝜋𝜋 𝑛𝑛)
𝑘𝑘 256 256
2
𝑥𝑥(𝑛𝑛) = � 𝐴𝐴𝑘𝑘 cos( 2𝜋𝜋 𝑛𝑛 + 𝜙𝜙𝑘𝑘 ) Large low frequency component
𝑁𝑁 Small high frequency component
𝑘𝑘=0 𝐴𝐴0 = 1
𝐴𝐴1 = 0.8 1

𝐴𝐴20 = 0.1
• A0 controls the average level,
also called the DC offset. 0
0 50 100 150 200 250
n
• Ak controls the amplitude of
the cosine with frequency k. 1 + 0.1 cos( 2𝜋𝜋
1
𝑛𝑛) + 0.8 cos( 2𝜋𝜋
20
𝑛𝑛)
256 256
• φk controls the phase of the 2

cosine with frequency k. 𝐴𝐴0 = 1

𝐴𝐴1 = 0.1 1
𝐴𝐴20 = 0.8

0
0 50 100 150 200 250
n

Small low frequency component

(𝜙𝜙𝑘𝑘 = 0 for all 𝑘𝑘) Large high frequency component
ELEC1200 16
 Amplitude, Phase, and Frequency
• Amplitude Ak
– Tells us how much of each frequency there is.
– A0 controls the average level of the signal and is sometimes called the DC
component.
• Phase φk
– Not so important, just shifts each cosine left or right.
• Frequency k
– Tells us about how quickly the signal changes
– Low frequencies (small k) indicate slow changes.
– High frequencies (large k) indicate fast changes.
• We are most interested in the amplitude and how it changes with k.
– The frequency components with the largest amplitude are the “most important.”

ELEC1200 17
 Amplitude Spectrum
• A plot of Ak versus frequency k
• Gives a graphical representation of which frequencies are most
important
Large low frequency component Large high frequency component
Small high frequency component Small low frequency component
2 2
𝑥𝑥(𝑛𝑛) 𝑥𝑥(𝑛𝑛)
1 1

0
0 50 100 150 200 250
𝑛𝑛 0 𝑛𝑛
0 50 100 150 200 250

1 1
𝐴𝐴𝑘𝑘 𝐴𝐴0 = 1 𝐴𝐴0 = 1
𝐴𝐴1 = 0.8
𝐴𝐴𝑘𝑘
0.5 0.5 𝐴𝐴1 = 0.1
𝐴𝐴20 = 0.1 𝐴𝐴20 = 0.8
0
0 10 20 30 40
𝑘𝑘 0
0 10 20 30 40
𝑘𝑘

ELEC1200 18
 Example: Square Wave Tone

Waveform Amplitude Spectrum

1.5
1 x(n)
1
0 2
A2cos(2π64n+φ2)
0.5
-1
n 0 k
0 16 32 48 64 0 8 16 24 32
1.5 x(n)
1
1
0 2
A2cos(2π 64n+φ2)
0.5 6
-1 +A6cos(2π 64n+φ6)
n 0 k
0 16 32 48 64 0 8 16 24 32
1.5
1 x(n)
1
0 2
A2cos(2π 64n+φ2)
0.5
-1 6
+A6cos(2π 64n+φ6)
n 0 k 10
0 16 32 48 64 0 8 16 24 32 +A10cos(2π 64 n+φ10)
ELEC1200 k 19
 More Complex Example
signal

-1
n
0 50 100 150 200 250

amplitude spectrum

0.2

0.1

0 k
0 20 40 60 80 100 120

ELEC1200 20
 Transforms

• The Fourier Series is only one of a set of representations

(transforms) you will study later:
– Fourier Transform
– Laplace Transform
– Z Transform
• Transforms are merely a different way of expressing the same data.
No information is lost or gained when taking a transform.
– Loose analogy: Transform ~ Translation
• We use transforms because
– It gives us a different way of viewing or understanding the signal.
– Some operations on signals are easier to understand/analyze after taking the
transform.

ELEC1200 21
 Summary
• Any signal can be expressed as the sum of cosines with different
amplitudes, frequencies and phases.

• The amplitude spectrum

– Is the weightings (sizes) of the different frequency components, Ak, plotted as
a function of frequency k.
– Tells us important information about what the signal looks like.
– Frequency components with the largest amplitudes are the “most important”.

• Knowledge of both amplitude and phase spectrum is equivalent to

knowledge of the signal.

ELEC1200 22
 Appendix: Discrete Fourier Transform (DFT)
• The DFT takes a waveform x(n) and returns a set of N complex valued
coefficients, called Fourier coefficients.
– The Fourier coefficients are denoted by Xk for k=0,1…(N-1)
– The formula (not important for this course) is
N −1

∑
− j 2 πkn
=Xk x(n)e =
N
for k 0,1,  , (N − 1)
n=0
This is the complex exponential
For more details wait until you take ELEC2100

– For this course, we use MATLAB’s “fft” function to compute the Fourier coefficients.

• From the first N/2+1 Fourier coefficients, we can compute Ak and φk for k=0,1,…,
(N/2)

 N1 Xk k = 0 or N
2
Ak =  2 φk =∠Xk
 N Xk otherwise

ELEC1200 23
 Appendix: complex numbers
• Define j= −1
• A complex number is given by z = a+jb, where
– a is the “real” part = Re{z}
– b is the “imaginary” part = Im{z}
• Since a complex number has two parts, we can represent it
as a point on a 2D plane (the complex plane)
• By analogy with polar coordinates, we can define the
magnitude and phase.
Im z

z
= a2 + b2 z= a + jb
b
∠z =arctan(b / a)
=a z cos(∠z) z
=b z sin(∠z)
∠z
a
ELEC1200
Re z 24
 Appendix: The DFT in MATLAB
• The MATLAB function “fft” implements the “Fast Fourier Transform,” a fast way
to compute the DFT.
>> Xdft = fft(x);
returns an N dimensional vector containing the DFT coefficients.
• We need to be a bit careful with MATLAB’s indexing
– The first index of x(n) in our notation is n=0, but the first index of a vector in
MATLAB starts with 1.
– Thus, if we represent a waveform x(n) for n=0,..(N-1) in a MATLAB vector x. Then,

x(1)= x(0), x(2)= x(1),…, x(N)= x(N-1)

MATLAB signal
– Similarly,
Xdft(1) = X0

ELEC1200 25