2 Signals and Signal Space

▪
▪
Size of a signal
Classification of signals
▪ Some useful signal operations
▪ Unit step and unit impulse signals

▪ Trigonometric Fourier series

▪ The exponential Fourier series
▪ Orthogonal signal sets

▪ Signals versus vectors

▪ Correlation of signals
▪ Matlab exercises
2

Classification of Signals

 Continuous time and discrete time

 Analog and digital

 Periodic and aperiodic

 Energy and power

 Deterministic and probabilistic

(represented completely by a mathematical representation)

3

Continuous-time and discrete-time signals

 g 𝑡 is a continuous-
time signal if 𝑡 is
continuous

 g[𝑛] is discrete-time
signal if 𝑛 is discrete
4

Analog and digital signals

Analog and Digital and

continuous- continuous-
time time

Analog and Digital and

discrete- discrete-
time time
5

Periodic and aperiodic signals

Periodic signal (from −∞ 𝑡𝑜 + ∞) and satisfy:

𝑔 𝑡 = 𝑔(𝑡 + 𝑇𝑜 )

Where 𝑇𝑜 is the period

6

Basics..
Amplitude (volt, Ampere) Peak value = ?
Power = ? Energy= ?

RMS = ?

Average value
(DC part)=?

Frequency(f) =?

Period (T)=?

Duration (time)
7

From the figure:

𝐴𝑝𝑒𝑎𝑘 = 20 𝑉𝑝𝑒𝑎𝑘 = 20 𝑣𝑜𝑙𝑡, 𝐼𝑝𝑒𝑎𝑘 = 20 𝐴𝑚𝑝𝑒𝑟𝑒

𝐸𝑛𝑒𝑟𝑔𝑦, 𝐸 = ∞

𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑣𝑎𝑙𝑢𝑒 = 𝐷𝑐 𝑝𝑎𝑟𝑡 = 5

𝑃𝑒𝑟𝑖𝑜𝑑 𝑇𝑜 = 25 𝑚𝑠

1
𝑓= 𝐾𝐻𝑧 = 40𝐻𝑧
25
8

Energy and power signals

A signal with finite energy is an energy

signal
∞

න 𝑔 𝑡 𝑑𝑡 < ∞
−∞

A signal with finite power is power

signal
∞

න 𝑔 𝑡 𝑑𝑡 = ∞
−∞
Comments:
 All observed signal in real life are energy signals. Power signals require infinite
energy
 Not all power signals are periodic
9

Signal Energy
For a signal 𝑔 𝑡 : the total amount of its energy is
∞

𝐸𝑔 = න 𝑔2 𝑡 𝑑𝑡 joule
−∞

Or more generally for complex signals:

𝐸𝑔 = න 𝑔(𝑡) 2 𝑑𝑡 joule
−∞
10

Signal Power

For a signal 𝑔 𝑡 : the average value of its power is the

square of the signal rms as:
1 𝑇/2 2
𝑃𝑔 = lim න 𝑔 𝑡 𝑑𝑡 watt
𝑇→∞ 𝑇 −𝑇/2

Or more generally for complex signals:

1 𝑇/2 2
𝑃𝑔 = lim න 𝑔(𝑡) 𝑑𝑡 watt
𝑇→∞ 𝑇 −𝑇/2
11

Another Units of Signal Power

Logarithmic scales (easy to deal with large & small numbers)

10 log(𝑃) 𝑑𝐵𝑤 Or 30 + 10 log(𝑃) 𝑑𝐵𝑚

Watt dBw dBm

10−6 = 0.0000001 -60 -30
10−3 = 1 𝑚𝑊 -30 0
1 0 30
1000 000 60 90

Note: How much is 50 dBw to 40 dBw?

(10 times)
12

Determine the suitable measure (energy or power) for the shown

two signals:

Answer: The first one is and energy signal and the second one is a
power signal
∞ 0 ∞

𝐸𝑔 = න 𝑔2 𝑡 𝑑𝑡 = න 22 𝑡 𝑑𝑡 + න 4𝑒 −𝑡 𝑑𝑡 2𝑒 −𝑡/2

−∞ −1 0
= 4 + 4 = 8 joule

1 1
1 2
1 2
1
𝑃𝑔 = න 𝑔 𝑡 𝑑𝑡 = න 𝑡 𝑑𝑡 = 𝑤𝑎𝑡𝑡
2 2 3
−1 −1
13

RMS vs Power
Determine the rms and power for the following signals:
a) 𝑔 𝑡 = 𝐶 cos 𝜔𝑜 𝑡 + 𝜃
𝑇/2
1
𝑃𝑔 = lim න 𝐶 2 𝑐𝑜𝑠 2 𝜔𝑜 𝑡 + 𝜃 𝑑𝑡
𝑇→∞ 𝑇
−𝑇/2
𝑇/2
1 𝐶2
= lim න 1 + cos 2𝜔𝑜 𝑡 + 2𝜃 𝑑𝑡
𝑇→∞ 𝑇 2
−𝑇/2 0
𝑇/2 𝑇/2
2
𝐶 𝐶2
= lim න 𝑑𝑡 + lim න cos 2𝜔𝑜 𝑡 + 2𝜃 𝑑𝑡
𝑇→∞ 2𝑇 𝑇→∞ 2𝑇
−𝑇/2 −𝑇/2
𝐶 2 𝑇 −𝑇 𝐶2
= − = 𝑤𝑎𝑡𝑡
2𝑇 2 2 2
The rms value is 𝐶/ 2
14

RMS vs Power
b) 𝑔 𝑡 = 𝐶1 cos 𝜔1 𝑡 + 𝜃 + 𝐶2 cos 𝜔2 𝑡 + 𝜃
Like previous example

𝐶12 𝐶22
𝑃𝑔 = +
2 2

The rms value is (𝐶12 + 𝐶22 )/2

15

Useful signal time-operations

1− Time Reversal (Reflection)

𝑥(−𝑡) is a reversal signal of 𝑥(𝑡)

16

2−Time Shift

𝑥(𝑡 + 𝑡𝑜 ) is a shifted version of the signal to the left

17

3−Time Scaling

If 𝑥 𝑡 =

Then : 𝑥 2𝑡 =

And : 𝑥 𝑡/2 =
18

Example:
My way for drawing:
For the shown 𝑥(𝑡) 1. Sign
2. Scale
3. Shift (for 𝑡 only)

Find the following:

 𝑥(𝑡 + 1)
19

 𝑥 −𝑡 + 1
My way for drawing:
= 𝑥(− 𝑡 − 1 ) 1. Sign
2. Scale
3. Shift (for 𝑡 only)

3
 𝑥 𝑡
2
20

3
 𝑥 𝑡+1
2 My way for drawing:
3 2
= 𝑥 (𝑡 + ) 1. Sign
2 3
2. Scale
3. Shift (for 𝑡 only)

−3
 𝑥 𝑡+1
2
−3 2
= 𝑥 (𝑡 − )
2 3
1- reverse the original 𝑥 𝑡 signal because of the (-);
2
2- scale the new signal time by
3
2
3- shift the resultant to right by
3
21

(Another method):
For 𝑥(𝑡)

3
To plot 𝑥 𝑡 + 1 (for example)
2
3 2
What was on 𝑡𝑜𝑙𝑑 = 2 will be in: 𝑡 +1 = 2 → 𝑡𝑛𝑒𝑤 =
2 𝑛𝑒𝑤 3
3
What was on 𝑡𝑜𝑙𝑑 = 1 will be in: 𝑡 +1 = 1 → 𝑡𝑛𝑒𝑤 = 0
2 𝑛𝑒𝑤
3 −2
What was on 𝑡𝑜𝑙𝑑 = 0 will be in: 𝑡 +1 = 0 → 𝑡𝑛𝑒𝑤 =
2 𝑛𝑒𝑤 3
22

Unit Step (U) Function

1, 𝑡>0
𝑢 𝑡 =ቊ
0, 𝑡<0
23

Unit impulse ( 𝛿 ) signal

1 ,𝑡 = 0
𝛿(𝑡) = ቊ
0 ,𝑡 ≠ 0

𝑡
based on න 𝛿 𝑡 𝑑𝑡 = 𝑢(𝑡) The 𝜹 function is a strange function that raises
and again to rest again in no time!
−∞
∞
𝑑
න 𝛿(𝑡) 𝑑𝑡 = 1 𝛿 𝑡 = 𝑢 𝑡
𝑑𝑡
−∞
24

∅ 𝑡 𝛿 𝑡 =∅ 0 𝛿 𝑡

∅ 𝑡 𝛿 𝑡−𝑇 =∅ 𝑇 𝛿 𝑡−𝑇

Sampling (sifting) property of the unit impulse function:

∞ ∞

න ∅ 𝑇 𝛿 𝑡 − 𝑇 𝑑𝑡 = ∅ 𝑇 න 𝛿 𝑡 − 𝑇 𝑑𝑡 = ∅ 𝑇
−∞ −∞

𝑏
−∅ 𝑇 , 𝑎≤𝑇≤𝑏
න ∅ 𝑇 𝛿 𝑡 − 𝑇 𝑑𝑡 = ቊ
0, 𝑒𝑙𝑒𝑠𝑒𝑤ℎ𝑒𝑟𝑒
𝑎
25

▪ Fourier Series vs Fourier Transform

▪ Trigonometric Fourier series

▪ Exponential Fourier series

26

 What is the Fourier Series/Transform for a signal?

 Why orthogonal signals?

27

Fourier Series vs Fourier Transform

 The Fourier series is used to represent a periodic function by a

discrete sum

 The Fourier transform is used to represent a nonperiodic function

by a continuous superposition or integral of complex exponentials.

 The Fourier transform can be viewed as the limit of the Fourier

series of a function with the period approaches to infinity (𝑇𝑜 → ∞)
28

4 4 4 4 4
sin 𝑡 + sin 3𝑡 + sin 5𝑡 + sin 7𝑡 + ⋯ + sin(𝑛𝑡)
𝜋 3𝜋 5𝜋 7𝜋 𝑛𝜋
29
30

 Executing a Fourier series for a periodic signal 𝑔෤ 𝑡 would result in

a signal 𝜑 𝑡

 Both: 𝑔෤ 𝑡 and 𝜑 𝑡 are in time domain.

 𝑔 𝑡 is a one period signal extracted from the periodic 𝑔෤ 𝑡

 Fortunately, executing the Fourier series for the aperiodic 𝑔 𝑡

results in the same 𝜑 𝑡 [Proof will be in next slides]
Note: this is because the original signal 𝑔෤ 𝑡 is periodic. This rule is
not applicable for an original aperiodic signals.
31

Trigonometric Fourier series

∞

𝜑 𝑡 = 𝑔෤ 𝑡 = 𝑎0 + ෍ 𝑎𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡) + 𝑏𝑛 𝑠𝑖𝑛(𝑛𝜔0 𝑡)
𝑛=1
Where
2𝜋
2 𝜔0 = 2𝜋𝑓0 =
𝑇0
𝑎𝑛 = න 𝑔 𝑡 𝑐𝑜𝑠(𝑛𝜔0 𝑡)𝑑𝑡
𝑇0
𝑇0
1 𝜔 = 2𝜋𝑛𝑓0
Special case 𝑎𝑜 = 𝑇 ‫𝑇׬‬0 𝑔(𝑡)𝑑𝑡
0

2
𝑏𝑛 = න 𝑔 𝑡 𝑠𝑖𝑛(𝑛𝜔0 𝑡)𝑑𝑡
𝑇0
𝑇0

𝜑 𝑡 = 𝑎0 + 𝑎1 𝑐𝑜𝑠(𝜔0 𝑡)+ 𝑎2 𝑐𝑜𝑠(2𝜔0 𝑡)+ ⋯ .

+𝑏1 𝑠𝑖𝑛 𝜔0 𝑡 + 𝑏2 𝑠𝑖𝑛 2𝜔0 𝑡 + ⋯ .
32

Special cases:
 Even signal (only cos’s)
∞

𝜑 𝑡 = 𝑎0 + ෍ 𝑎𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡)
𝑛=1

 Odd signal (only sin’s)

∞

𝜑 𝑡 = 𝑎0 + ෍ 𝑏𝑛 𝑠𝑖𝑛(𝑛𝜔0 𝑡)
𝑛=1
33

Special Case II
Compact form: If 𝜑 𝑡 is real periodic signal (no complex
component), then it can be represented as:
∞

𝜑 𝑡 = 𝑔෤ 𝑡 = 𝐶0 + ෍ 𝐶𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡 + 𝜃)
𝑛=1
Where:
𝐶0 = 𝑎0
𝐶𝑛 = 𝑎𝑛2 + 𝑏𝑛2

−1
−𝑏𝑛
𝜃 = 𝑡𝑎𝑛
𝑎𝑛
34

From previous slides:

“ executing the Fourier series for the aperiodic 𝑔 𝑡 results in the
same 𝜑 𝑡 ”
The simplest proof is based on the compact form representation.

For a periodic signal 𝑔෤ 𝑡 with period 𝑇𝑜 , if the Fourier series was

calculated over the period 𝑇𝑜 only :
∞

𝜑 𝑡 = 𝐶0 + ෍ 𝐶𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡 + 𝜃) 𝐛𝐮𝐭 for 0 < 𝑡 < 𝑇𝑜 ,

𝑛=1
We want to check how 𝜑 𝑡 + 𝑇𝑜 will looks like
35

Cont..
∞

𝜑 𝑡 = 𝐶0 + ෍ 𝐶𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡 + 𝜃) 0 < 𝑡 < 𝑇𝑜

𝑛=1
∞

𝜑 𝑡 + 𝑇𝑜 = 𝐶0 + ෍ 𝐶𝑛 𝑐𝑜𝑠(𝑛𝜔0 (𝑡 + 𝑇0 ) + 𝜃)
𝑛=1
But 𝑐𝑜𝑠 𝑛𝜔0 𝑡 + 𝑇0 = cos(𝑛𝜔0 𝑡 + 𝑛𝜔0 𝑇0 )
2𝜋
= cos 𝑛𝜔0 𝑡 + 𝑛 𝑇 = cos(𝑛𝜔0 𝑡 + 2𝜋𝑛) = cos(𝑛𝜔0 𝑡)
𝑇0 0

∴ 𝜑 𝑡 + 𝑇𝑜 = 𝐶0 + ෍ 𝐶𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡 + 𝜃)
𝑛=1
= 𝜑 𝑡 for all 𝑡 ∴ the statement is true
36

Example: Find the compact trigonometric Fourier series

for the signal 𝜑 𝑡 where 𝑔 𝑡 = 𝑒 −𝑡/2 shown in the Fig.

Note: check 𝑔 𝑡 and 𝜑 𝑡 on the plot

37

1 1
𝑇𝑜 = 𝜋 (from the graph) → 𝑓𝑜 = 𝐻𝑧 , 𝜔𝑜 = 2𝜋 = 2 𝑟𝑎𝑑/𝑠
𝜋 𝜋
∞
Spend some
𝜑 𝑡 = 𝑔෤ 𝑡 = 𝑎0 + ෍ 𝑎𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡) + 𝑏𝑛 𝑠𝑖𝑛(𝑛𝜔0 𝑡)
time to practice
𝑛=1
∞ these integrals
= 𝑎0 + ෍ 𝑎𝑛 𝑐𝑜𝑠(2𝑛𝑡) + 𝑏𝑛 𝑠𝑖𝑛(2𝑛𝑡)
𝑛=1
𝜋
1
𝑎𝑜 = න 𝑒 −𝑡/2 𝑑𝑡 = 0.504 1
𝜋 𝑎𝑜 = න 𝑔(𝑡)𝑑𝑡
0
𝜋 𝑇0
𝑇0
2 −
𝑡 2
𝑎𝑛 = න 𝑒 2 𝑐𝑜𝑠 2𝑛𝑡 𝑑𝑡 = 0.504 2
𝜋 1 + 16𝑛2 𝑎𝑛 = න 𝑔 𝑡 𝑐𝑜𝑠(𝑛𝜔0 𝑡)𝑑𝑡
0
𝜋 𝑇0
2 −
𝑡 8𝑛 𝑇0
𝑏𝑛 = න 𝑒 2 𝑠𝑖𝑛 2𝑛𝑡 𝑑𝑡 = 0.504
𝜋 1 + 16𝑛2 2
0 𝑏𝑛 = න 𝑔 𝑡 𝑠𝑖𝑛(𝑛𝜔0 𝑡)𝑑𝑡
𝑇0
𝑇0

2
∴ 𝜑 𝑡 = 𝑔෤ 𝑡 = 0.504 1 + σ∞
𝑛=1 1+16𝑛2 𝑐𝑜𝑠(2𝑛𝑡) + 4𝑛 𝑠𝑖𝑛(2𝑛𝑡) for any t
38

For the compact form: (if required)

𝐶𝑜 = 𝑎𝑜 = 0.504
4 64𝑛2 2
𝐶𝑛 = 𝑎𝑛2 + 𝑏𝑛2 = 0.504 + = 0.504
1 + 16𝑛2 2 1 + 16𝑛2 2
1 + 16𝑛2
−1
−𝑏𝑛
𝜃 = 𝑡𝑎𝑛 = 𝑡𝑎𝑛−1 −4𝑛 = −𝑡𝑎𝑛−1 4𝑛
𝑎𝑛
∞
2 −1 4𝑛
∴ 𝜑 𝑡 = 0.504 + 0.504 ෍ 𝑐𝑜𝑠(2𝑛𝑡 − 𝑡𝑎𝑛
1 + 16𝑛2
𝑛=1
= 0.504 + 0.244 𝑐𝑜𝑠 2𝑡 − 75.96𝑜 + 0.125 𝑐𝑜𝑠 4𝑡 − 82.87𝑜

+0.084 cos 6𝑡 − 85.24𝑜 + 0.063 cos 8𝑡 − 86.42𝑜 + ⋯

39

Plotting the Fourier series representation

 While the signal 𝜑 𝑡 is still in time domain, it is possible to represent the signal
in a frequency representation .
 Example: (from the previous example)
∞
2
𝜑 𝑡 = 0.504 + 0.504 ෍ 2
𝑐𝑜𝑠(2𝑛𝑡 − 𝑡𝑎𝑛−1 4𝑛
1 + 16𝑛
𝑛=1
= 0.504 + 0.244 𝑐𝑜𝑠 2𝑡 − 75.96𝑜 + 0.125 𝑐𝑜𝑠 4𝑡 − 82.87𝑜

+0.084 cos 6𝑡 − 85.24𝑜 + 0.063 cos 8𝑡 − 86.42𝑜 + ⋯

𝒏 0 1 2 3 4 5 6 7
𝐶𝑛 0.504 0.244 0.125 0.084 0.063 0.050 0.042 0.036
𝜃𝑛 0 −75.96𝑜 −82.87𝑜 −85.24𝑜 −86.42𝑜 −87.14𝑜 −87.61𝑜 −87.95𝑜
40

𝒏 0 1 2 3 4 5 6 7
𝐶𝑛 0.504 0.244 0.125 0.084 0.063 0.050 0.042 0.036
𝜃𝑛 0 −75.96𝑜 −82.87𝑜 −85.24𝑜 −86.42𝑜 −87.14𝑜 −87.61𝑜 −87.95𝑜
41

Complex Exponential Fourier Series

∞ 2𝜋
𝜔0 = 2𝜋𝑓0 =
𝜑 𝑡 = 𝑔෤ 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0𝑡 𝑇0

𝑛=−∞
Where: 𝜔 = 2𝜋𝑛𝑓0
1
𝐷𝑛 = න 𝑔(𝑡)𝑒 −𝑗𝑛𝜔0 𝑡 𝑑𝑡 𝑎𝑛 =Complex Fourier
𝑇0
𝑇0 Coefficient
1
𝐷𝑜 = න 𝑔(𝑡)𝑑𝑡 Every frequency parts
𝑇0
𝑇0 called harmonic

𝜑 𝑡 = ⋯ + 𝑎−2 𝑒 −2𝑗.. + 𝑎−1 𝑒 −𝑗.. + 𝑎0 + 𝑎1 𝑒 +𝑗.. + 𝑎2 𝑒 +2𝑗.. + ⋯

42

Example: Find the exponential Fourier series for the signal 𝜑 𝑡 where
𝑔 𝑡 = 𝑒 −𝑡/2 as shown in the Fig.

1 1
𝑇𝑜 = 𝜋 (from the graph) → 𝑓𝑜 = 𝐻𝑧 , 𝜔𝑜 = 2𝜋 = 2 𝑟𝑎𝑑/𝑠
𝜋 𝜋
43

𝜋
1 −𝑗𝑛𝜔0 𝑡
1 −𝑡/2 −𝑗𝑛𝜔0 𝑡
1 1
−( +𝑗𝑛𝜔0 )𝑡
𝐷𝑛 = න 𝑔(𝑡)𝑒 𝑑𝑡 = න 𝑒 𝑒 𝑑𝑡 = න 𝑒 2 𝑑𝑡
𝑇0 𝜋 𝜋
𝑇0 0 𝑇0
𝜋
−1 1
−( +𝑗𝑛𝜔0 )𝑡 ቚ 0.504
= 𝑒 2 0 =
1 1 + 𝑗4𝑛
𝜋( + 𝑗𝑛𝜔0 )
2
Apply to:
∞ ∞
1
𝜑 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0 𝑡 = 0.504 ෍ 𝑒 𝑗2𝑛𝑡
1 + 𝑗4𝑛
𝑛=−∞ 𝑛=−∞

1 𝑗2𝑡
1 𝑗4𝑡
1 −𝑗2𝑡
1
= 0.504 1 + 𝑒 + 𝑒 +⋯ + 𝑒 + 𝑒 −𝑗4𝑡 + ⋯
1 + 𝑗4 1 + 𝑗8 1 − 𝑗4 1 − 𝑗8
44

To plot:

1 1 1 1
𝜑 𝑡 = 0.504 1 + 𝑒 𝑗2𝑡 + 𝑒 𝑗4𝑡 + ⋯ + 𝑒 −𝑗2𝑡 + 𝑒 −𝑗4𝑡 + ⋯
1 + 𝑗4 1 + 𝑗8 1 − 𝑗4 1 − 𝑗8

𝐷0 = 0.504 (DC part)

0.504
𝐷1 = = 0.122𝑒 −𝑗75.96 = 𝐷𝑛 ∠𝜃 = 0.122∠ − 75.96
1+𝑗4
0.504
𝐷−1 = = 0.122𝑒 −𝑗75.96 = 0.122∠75.96
1−𝑗4
0.504
𝐷2 = = 0.122𝑒 −𝑗75.96 = 0.0625∠ − 82.87
1+𝑗8
0.504
𝐷−2 = = 0.122𝑒 −𝑗75.96 = 0.0625∠82.87
1−𝑗8
….
45

𝐷0 = 0.504
𝐷1 = 0.122∠ − 75.96
𝐷−1 = 0.122∠75.96
𝐷2 = 0.0625∠ − 82.87
𝐷−2 = 0.0625∠82.87
…..

= 90𝑜
46

Example: Find the Fourier series representation for the

shown periodic signal 𝑔(𝑡)
෤ knowing that 𝑇𝑜 = 2,
𝑔(𝑡)

𝑡
−𝜏/2 −𝜏/4 𝜏/4 𝜏/2

2𝜋 1𝜋 𝑇𝑜
𝜔𝑜 = = = 𝜋 𝑟𝑎𝑑/𝑠
𝑇𝑜 2
1 1
𝑓𝑜 = = 𝐻𝑧
𝑇𝑜 2
47

1/2
1 1 1 1/2 1 1
𝐷𝑜 = න 𝑔 𝑡 𝑑𝑡 = න 1 𝑑𝑡 = 𝑡ቚ = 1 =
𝑇0 2 2 −1/2 2 2
𝑇0 −1/2

1/2
1 1
𝐷𝑛 = න 𝑔 𝑡 𝑒 −𝑗𝑛𝜔0𝑡 𝑑𝑡 = න 1 𝑒 −𝑗𝑛𝜋𝑡 𝑑𝑡
𝑇0 2
𝑇0 −1/2
1/2 1/2
−𝑗𝑛𝜋𝑡 −𝑗𝑛𝜋𝑡
1 𝑒 1 1 𝑒
= อ = อ
2 −𝑗𝑛𝜋 2 𝑛𝜋 −𝑗
−1/2 −1/2

1 𝑒 𝑗𝑛𝜋/2 − 𝑒 −𝑗𝑛𝜋/2 1 𝑘𝜋 Harmonics

= 2 = 𝑠𝑖𝑛 of multiple of
2𝑛𝜋 2𝑗 𝑛𝜋 2
1/4
1 1
Knowing that ω = 2𝜋𝑓, rewrite the above as: 𝑠𝑖𝑛 2𝜋𝑘
𝑛𝜋 4
48

𝐷0 = 1/2
𝐷𝑛
Based on: 1 1 1
1 𝑘𝜋 2
𝐷𝑛 = 𝑠𝑖𝑛 1 𝜋 𝜋 1
𝑛𝜋 2 1 1
3𝜋 3𝜋
5𝜋 5𝜋
1
𝐷1 = 𝐷−1 = ∠0
𝜋 1 1 3 5 3
0 1 𝑓
𝐷2 = 𝐷−2 = 0 4 2 4 4 2
1 ∠𝐷𝑛
𝐷3 = 𝐷−3 = ∠𝜋 𝜋 𝜋 𝜋 𝜋 𝜋 𝜋 𝜋 𝜋 𝜋
3𝜋
𝐷4 = 𝐷−4 = 0
1
𝐷5 = 𝐷−5 = ∠0
5𝜋 1 1 3 5 3
𝐷6 = 𝐷−6 = 0 0 1 𝑓
4 2 4 4 2
1
𝐷7 = 𝐷−7 = ∠𝜋
7𝜋
49
50

Example: Find the Fourier series representation for the signal 𝑔෤ 𝑡 = 𝑔(𝑡 + 2𝑛𝜋)
where 𝑔(𝑡) is as shown below. Find both representations: the trigonometric and
exponential form

𝑡 0<𝑡≤𝜋
𝑔 𝑡 =ቊ
𝜋 𝜋 < 𝑡 ≤ 2𝜋

Solution 1:
2𝜋
𝑇𝑜 = 2𝜋 (from the question), 𝜔𝑜 = 𝑇𝑜
=1

2𝜋 𝜋 2𝜋 𝜋
1 1 1 𝑡2 2𝜋
𝑎𝑜 = න 𝑔(𝑡)𝑑𝑡 = න 𝑡 𝑑𝑡 + න 𝜋 𝑑𝑡 = อ + 𝜋𝑡ቚ
𝑇0 2𝜋 2𝜋 2 𝜋
0 0 𝜋 0
2
1 𝜋 3
= − 0 + 2𝜋 2 − 𝜋 2 = 𝜋
2𝜋 2 4
51

2𝜋
2 2
𝑎𝑛 = න 𝑔 𝑡 𝑐𝑜𝑠(𝑛𝑡)𝑑𝑡 = න 𝑔 𝑡 𝑐𝑜𝑠(𝑛𝑡)𝑑𝑡
𝑇0 2𝜋
𝑇0 0

𝜋 2𝜋
1 The cos integration is some how
= න 𝑡 𝑐𝑜𝑠 𝑛𝑡 𝑑𝑡 + න 𝜋 𝑐𝑜𝑠 𝑛𝑡 𝑑𝑡
𝜋 more difficult than the exp
0 𝜋

𝜋 2𝜋
1 𝑐𝑜𝑠 𝑛𝑡 + 𝑛 𝑡 𝑠𝑖𝑛(𝑛𝑡) −𝜋 𝑐𝑜𝑠(𝑛𝑡)
= ቤ + ቤ
𝜋 𝑛2 0
𝑛 𝜋

1 𝑐𝑜𝑠 𝑛𝜋 + 𝑛𝜋 𝑠𝑖𝑛(𝑛𝜋) 1 −𝜋 𝑐𝑜𝑠(2𝜋𝑛) −𝜋 𝑐𝑜𝑠(𝜋𝑛)

= − 2 + −
𝜋 𝑛2 𝑛 𝑛 𝑛
No need for any further simplification
52

3
𝑎0 = 𝜋
4

1 𝑐𝑜𝑠 𝑛𝜋 + 𝑛𝜋 𝑠𝑖𝑛(𝑛𝜋) 1 −𝜋 𝑐𝑜𝑠(2𝜋𝑛) −𝜋 𝑐𝑜𝑠(𝜋𝑛)

𝑎𝑛 = − 2 + −
𝜋 𝑛2 𝑛 𝑛 𝑛
−2
𝑎1 =
𝜋
𝑎2 = 0
−2
𝑎3 =
9𝜋
𝑎4 = 0
−2
𝑎5 =
25𝜋
53

2𝜋
2 2
𝑏𝑛 = න 𝑔 𝑡 𝑠𝑖𝑛(𝑛𝑡)𝑑𝑡 = න 𝑔 𝑡 𝑠𝑖𝑛(𝑛𝑡)𝑑𝑡
𝑇0 2𝜋
𝑇0 0
𝜋 2𝜋
1
= න 𝑡 𝑠𝑖𝑛 𝑛𝑡 𝑑𝑡 + න 𝜋 sin 𝑛𝑡 𝑑𝑡 = ⋯
𝜋
0 𝜋

1 𝑠𝑖𝑛 𝑛𝜋 − 𝑛𝜋 𝑐𝑜𝑠(𝑛𝜋) −𝜋 𝑐𝑜𝑠(2𝜋𝑛) −𝜋 𝑐𝑜𝑠(𝜋𝑛)

= 2
−0 + −
𝜋 𝑛 𝑛 𝑛

1 −1 −1
𝑏1 = −1 𝑏2 = − 𝑏3 = 𝑏4 =
2 3 4
54

∞
3
𝜑 𝑡 = 𝑔෤ 𝑡 = 𝑎0 + ෍ 𝑎𝑛 𝑐𝑜𝑠(𝑛𝜔0 𝑡) + 𝑏𝑛 𝑠𝑖𝑛(𝑛𝜔0 𝑡) 𝑎0 = 4 𝜋
𝑛=1
−2
𝑎1 = 𝜋
3 2 𝑎2 = 0
= 𝜋 − 𝑐𝑜𝑠 𝑡 − (1) 𝑠𝑖𝑛(𝑡)
4 𝜋 −2
𝑎3 = 9𝜋
1st harmonic 𝑎4 = 0
1
− 𝑠𝑖𝑛(2𝑡) 2nd harmonic
2
3rd harmonic 𝑏1 = −1
1
𝑏2 = − 2
2 1
− 𝑐𝑜𝑠 3𝑡 − 𝑠𝑖𝑛 3𝑡 𝑏3 =
−1
9𝜋 3 3
−1
𝑏4 = 4
+…
55

Solution II: Exponential representation From

previous
3 solution:
𝐷𝑜 = 𝑎𝑜 = 𝜋 3
4 𝑎0 = 𝜋
4
𝑇𝑜 = 2𝜋
2𝜋
𝜋 2𝜋 𝜔𝑜 =
𝑇𝑜
1 −𝑗𝑛𝜔 𝑡
1 =1
𝐷𝑛 = න 𝑔(𝑡)𝑒 0 𝑑𝑡 = න 𝑡 𝑒 −𝑗𝑛𝑡 𝑑𝑡 + න 𝜋𝑒 −𝑗𝑛𝑡 𝑑𝑡
𝑇0 2𝜋
𝑇0 0 𝜋
1 𝜋 2𝜋
= 1 ቚ + 2 ቚ
2𝜋 0 𝜋
For 1 , let 𝑢 = 𝑡 ⇒ 𝑑𝑢 = 𝑑𝑡 න 𝑢 𝑑𝑣 = 𝑢 𝑣 − 𝑣 𝑑𝑢

𝑒 −𝑗𝑛𝑡 𝑗 −𝑗𝑛𝑡
𝑑𝑣 = 𝑒 −𝑗𝑛𝑡 𝑑𝑡 ⇒ 𝑣 = = 𝑒
−𝑗𝑛 𝑛
56

𝑗 −𝑗𝑛𝑡 𝑗 −𝑗𝑛𝑡 𝑡 −𝑗𝑛𝑡 𝑗 𝑒 −𝑗𝑛𝑡 𝑡 1

∴ 1 =𝑡 𝑒 −න 𝑒 𝑑𝑡 = 𝑗 𝑒 − = 𝑗 𝑒 −𝑗𝑛𝑡 + 2 𝑒 −𝑗𝑛𝑡
𝑛 𝑛 𝑛 𝑛 −𝑗𝑛 𝑛 𝑛
𝜋
𝜋 𝑡 −𝑗𝑛𝑡 1 −𝑗𝑛𝑡 𝜋 −𝑗𝑛𝑡 1 −𝑗𝑛𝜋 1
∴ 1 ቚ = 𝑗 𝑒 + 2𝑒 อ =𝑗 𝑒 + 2𝑒 − 𝑗0 − 2
0 𝑛 𝑛 𝑛 𝑛 𝑛
0
2𝜋
For 2 ห𝜋 :
2𝜋 2𝜋
−𝑗𝑛𝑡
−𝑗𝑛𝑡
𝑒 𝜋 −𝑗2𝑛𝜋
න 𝜋𝑒 𝑑𝑡 = 𝜋 อ =𝑗 𝑒 − 𝑒 −𝑗𝑛𝜋
−𝑗𝑛 𝑛
𝜋 𝜋
1 𝜋 2𝜋
∴ 1 ቚ + 2 ቚ
2𝜋 0 𝜋
1 −𝑗𝑛𝜋 1 −𝑗𝑛𝜋
1 1 −𝑗2𝑛𝜋 1 −𝑗𝑛𝜋
=𝑗 𝑒 + 𝑒 − +𝑗 𝑒 −𝑗 𝑒
2𝑛 2𝜋𝑛2 2𝜋𝑛2 2𝑛 2𝑛

1 −𝑗𝑛𝜋
1 1 −𝑗2𝑛𝜋
= 𝑒 − +𝑗 𝑒
2𝜋𝑛2 2𝜋𝑛2 2𝑛
57

3
𝐷𝑜 = 𝜋 = 2.36 ∠0
4

1 1 1 −𝑗2𝑛𝜋
𝐷𝑛 = 2 𝑒 −𝑗𝑛𝜋 − +𝑗 𝑒
2𝜋𝑛 2𝜋𝑛2 2𝑛

−1 1 −1 1
𝐷1 = 𝜋
+ 𝑗2 = 0.59 ∠2.13 𝐷−1 = 𝜋
− 𝑗2 = 0.59 ∠ − 2.13

1 1
𝐷2 = 𝑗 4 = 0.25 ∠ 1.57 𝐷−2 = −𝑗 4 = 0.25 ∠ − 1.57

−1 1 −1 1
𝐷3 = + 𝑗6 = 0.17 ∠ 1.7 𝐷−3 = − 𝑗6 = 0.17 ∠ − 1.7
9𝜋 9𝜋

1 1
𝐷4 = 𝑗 8 = 0.13 ∠ 1.57 𝐷−4 = 𝑗 8 = 0.13 ∠ − 1.57
58

𝜑 𝑡 = 𝑔෤ 𝑡 = ෍ 𝐷𝑛 𝑒 𝑗𝑛𝜔0𝑡
𝑛=−∞
𝜑 𝑡 = 𝑔෤ 𝑡 =

1 −1 1 1 −1 1
… . + 𝑗 𝑒 −𝑗4𝑡 + − 𝑗 𝑒 −𝑗3𝑡 + −𝑗 𝑒 −𝑗2𝑡 + − 𝑗 𝑒 −𝑗𝑡
8 9𝜋 6 4 𝜋 2

3
+ 𝜋
4

−1 1 𝑗𝑡 1 𝑗2𝑡 −1 1 𝑗3𝑡 1 𝑗4𝑡

+ − 𝑗 𝑒 + −𝑗 𝑒 + − 𝑗 𝑒 + 𝑗 𝑒 + ⋯.
𝜋 2 4 9𝜋 6 8
59
60

Vectors vs Signals and the

correlation
Component of a vector along another vector
Decomposition of a signal and signal component
Correlation of signals
Correlation and auto correlation functions
61

Vectors vs Signals
 A (limited) discrete signal is just a set of vectors
 And a continuous signal is a discrete signal with very
small-time lapse
lim 𝑥 𝑛 = 𝑥(𝑡)
𝑛𝑚+1 −𝑛𝑚 →0

 Then the signal 𝑥 𝑡 = vector x

can be represented by another
vector g with an error e
 Minimizing e is a goal. And result
in the following two important
equations:
𝑐𝐱 = 𝐠 𝑐𝑜𝑠 (𝜃)
62

If 𝑔(𝑡) is approximated by another signal 𝑥 𝑡 as:

𝑔(𝑡) ≈ 𝑐𝑥(𝑡)
Then the optimum c is calculated as:

𝑡2 𝑡2
‫𝑥 𝑡 𝑔 𝑡׬‬ 𝑡 𝑑𝑡 1
1
𝑐= 𝑡2 2 = න 𝑔 𝑡 𝑥 𝑡 𝑑𝑡
‫𝑡 𝑥 𝑡׬‬ 𝑑𝑡 𝐸𝑥
1 𝑡1

See example 2.5 p58

63

Correlation of signals
Two vectors g and x are similar if g has a large component along x (or: if
c is large)

However, it is better to compare the signal by the angle (correlation

coefficient 𝜌) :
−1 ≤ 𝜌 ≤ 1
< 𝐠. 𝐱 >
𝜌 = cos 𝜃 =
𝐠 𝐱 • 𝜌=1
𝐱 is the magnitude (length) of vector x Max similarity
< 𝐠. 𝐱 > = 𝐠 . 𝐱 𝑐𝑜𝑠(𝜃) • 𝜌=0
∞ Orthogonal
1 • 𝜌 = −1
𝜌= න 𝑔 𝑡 𝑥 ∗ 𝑡 𝑑𝑡
𝐸𝑥 𝐸𝑔 Max dissimilarity
−∞
64

The Auto Correlation function:

It is the correlation of a signal with itself (shifted version
of the signal)
∞ Φ=Phi
𝜑=variant Phi
𝜓𝑔 𝜏 = න 𝑔 𝑡 𝑔 𝑡 + 𝜏 𝑑𝑡 Ψ=Psi
−∞ 𝜓=variant Psi

∞
Note: 𝑔 𝑡 ∗ 𝑔 𝑡 = ‫׬‬−∞ 𝑔 𝜏 𝑔 𝑡 − 𝜏 𝑑𝜏
What is the correlation in terms of convolution

Example 2.6 p42

65

Time Autocorrelation Function and Energy Spectral Density:

◦ The correlation between two signals: 𝑔 𝑡 and ℎ(𝑡) was defined as:
∞

𝜓 𝑡 = න 𝑔 𝜏 ℎ 𝑡 + 𝜏 𝑑𝜏 = 𝑔 𝑡 ∗ ℎ −𝑡
−∞

◦ And the auto correlation for a signal 𝑔 𝑡 as:

𝝍 𝒕 = 𝒈 𝒕 ∗ 𝒈 −𝒕 ⇔ 𝜳 𝒇 = 𝑮 𝒇 𝑮(−𝒇)

◦ The ESD for the autocorrelated signal 𝜓 𝑡 is simply the 𝜳 𝒇