1

EE4900/EE6420: Digital Communications

Lecture 6
Modulation,
Demodulation
(Detection): Part 1

EE4900/EE6420 Digital Communications Suketu Naik

 2
Block Diagrams of Communication System
 Digital Communication System

Informatio
n (sound, Transducer & Source Channel Tx RF
A/D Converter Encoder Modulator System
video, text, Encoder
data, …)

Channel

Output D/A Converter Source Channel Demodulator Rx RF

Signal and/or output Decoder Decoder System
transducer

EE4900/EE6420 Digital Communications Suketu Naik

 3
EE4900/EE6720: Digital Communications

Introduction

EE4900/EE6420 Digital Communications Suketu Naik

 4
Digital Baseband Modulation Schemes
 Baseband Modulation is done before the information
(bits) is sent out using a carrier signal
 Popular methods for low data-rate are,
 ASK: Amplitude Shift Keying
 OOK: On-Off Keying
 FSK: Frequency Shift Keying
ASK Signal

OOK Signal

FSK Signal

EE4900/EE6420 Digital Communications Suketu Naik

 5
Digital Baseband Modulation Schemes
 Baseband Modulation is done before the information
(bits) is sent out using a carrier signal

 Popular methods for high data-rate are,

 PCM: Pulse Code Modulation
 PAM: Pulse Amplitude Modulation
 PWM: Pulse Width Modulation
 PPM: Pulse Position Modulation
 BPSK: Binary Phase Shift Keying
 QPSK: Quadrature Phase Shift Keying
 O-QPSK: Offset Quadrature Phase Shift Keying
 QAM: Quadrature Amplitude Modulation

EE4900/EE6420 Digital Communications Suketu Naik

 6
Pulse Code Modulation (PCM)
 PCM is commonly used for digital audio (Windows PCs,
CDs, digital telephony)
 PCM: Analog signal (such as voice or music) is
converted to discrete samples and then quantized to bits
 After quantization, companding (compressing and
expanding) takes place: μ-law in North America and
Japan and A-law in Europe and rest of the world

EE4900/EE6420 Digital Communications Suketu Naik

 7
Pulse Code Modulation (PCM)
 PCM Example
Original Signal

PCM Signal

EE4900/EE6420 Digital Communications Suketu Naik

 8
Pulse Amplitude Modulation (PAM)
 PAM is used in Digital TV, 10GBase-T Ethernet, LED
driver, Controls
 PAM: Width is fixed, Amplitude varies

Original Signal

PAM Signal

EE4900/EE6420 Digital Communications Suketu Naik

 9
Pulse Width or Duration Modulation (PWM or PDM)
 PWM is used in Controls (motors, servos for robots,
power, audio, telecomm.),
 PWM: Position is fixed, Width varies

Original Signal

PWM Signal

EE4900/EE6420 Digital Communications Suketu Naik

 10
Pulse Position Modulation (PPM)
 PPM is used in non-coherent RF comm., contactless
smart card, RFID tags, radio control, optical comm.
 PPM: Width is fixed, Position varies

Original Signal

PPM Signal

EE4900/EE6420 Digital Communications Suketu Naik

 11
Binary Phase Shift Keying (BPSK)
 BPSK is commonly used in communications
 BPSK: Phase of the baseband signal changes from 0 deg
to 180 deg (2 transitions)=1 bit per symbol

Original Signal

BPSK Symbol

Q (out of phase)

Phase Diagram 1 0
I (in phase)

EE4900/EE6420 Digital Communications Suketu Naik

 12
Quadrature Phase Shift Keying (QPSK)
 QPSK (and O-QPSK) is commonly used in
communications
 QPSK: Phase of the baseband signal changes from 45
deg, 135 deg, 225 deg, 315 deg (4 transitions)=2 bits per
symbol
Original Signal

QPSK Symbol

Q (out of phase)
01 11

Phase Diagram
I (in phase)
00 10

EE4900/EE6420 Digital Communications Suketu Naik

 13
8-PSK
 8PSK = 8 phase transitions= 3 bits per symbol

Q (out of phase)
010
011 110

001 111
8-PSK Phase I (in phase)
Diagram

000 101

100

EE4900/EE6420 Digital Communications Suketu Naik

 14
Quadrature Amplitude Modulation (QAM)
 QAM (rhymes with Guam and VietNAM) is commonly
used in communications
 QAM: Amplitude and Phase of the Baseband signal
changes, 16-QAM (16 transitions, 4-bits/symbol), 32-QAM
(5-bits/symbol), 64-QAM (64 transitions, 6-bits/symbol)
Q (out of phase)
16-QAM
Phase Diagram

I (in phase)

EE4900/EE6420 Digital Communications Suketu Naik

 15
Quadrature Amplitude Modulation (QAM)
 QAM Signal in Time-domain

8-QAM=8PSK

EE4900/EE6420 Digital Communications Suketu Naik

 16
EE4900/EE6420: Digital Communications

Basics of
Modulation

EE4900/EE6420 Digital Communications Suketu Naik

 17
Waveforms=Bits
 M-ary (“emery”) digital communication system uses M
symbols to transmit log2(M) bits
Examples

 BPSK: M=2, log2(M)=1 bit per symbol

Two symbols, one for each bit (0 and 1)

 QPSK: M=4, log2(M)=2 bits per symbol

Four symbols, one for two bit pairs
(00, 01, 11, 10)

 Binary PAM: M=2, 1 bit per symbol

 8-PAM: M=8, 3 bits per symbol
 16-QAM, 32-QAM, 64-QAM, 128-QAM: ?

EE4900/EE6420 Digital Communications Suketu Naik

 18
Waveforms, Symbols, Basis Functions
 M-ary communication uses M symbols
 Symbols are created by changing amplitudes of basis
functions
Basis functions > Symbols > Waveform
 Thus a waveform is a linear combination or weighted
sum of basis functions

This waveform can be called the baseband signal or the

modulating signal

 The modulating signal modulates (multiplied with) the

carrier signal so that the carrier signal can now carry
“bits”
EE4900/EE6420 Digital Communications Suketu Naik
 19
Signal Space and Basis Functions
 Let’s represent the M waveforms as a signal (symbol) set
𝑺 = {𝒔𝟎 𝒕 , 𝒔𝟏 𝒕 , 𝒔𝟐 𝒕 , … , 𝒔𝑴−𝟏 𝒕 }
Example Baseband Signal s(t)
with symbol time Ts
s (t) s (t) s (t) s (t) s (t) s (t) s (t) s (t)
0 1 2 3 4 5 6 7
 B=set of basis functions
 Signal Space=Span{B}= linear
combination of basis functions
Ts 2Ts 3Ts 4Ts …
 Baseband signal s(t) (with K=M) Example
𝑲−𝟏 1) M=8 waveforms=8 signals (symbols)
2) Signal (symbol) set S contains 8
𝒔(𝒕) = ෍ 𝒂𝒌 ∅𝒌 (𝒕) possible signals (symbols)
𝒌=𝟎 3) Each symbol represents 3-bits
e.g. bit patterns such as 101 or 100 are
Amplitude Basis Function mapped to specific symbols
4) s(t), the baseband signal, is the
linear combination of the basis
𝑻𝟐 𝟐
 Energy 𝑬𝒌 = ‫𝒌∅ 𝑻׬‬ 𝒕 𝒅𝒕 functions, constructed from symbols,
𝟏 s0(t), …sM-1(t)
EE4900/EE6420 Digital Communications Suketu Naik
 20
Properties of Basis Functions
 Orthogonality
𝑻𝟐
න ∅𝒊 𝒕 ∅𝒋 𝒕 𝒅𝒕 = 𝟎
𝑻𝟏
Example: cos(2πt) sin(2πt)
Interpretation:
Multiply two different signals and integrate, the result is 0
 Orthonormal Set
𝑻𝟐
𝟏 𝒊𝒇 𝒊 = 𝒋
න ∅𝒊 𝒕 ∅𝒋 𝒕 𝒅𝒕 = 𝜹 𝒊 − 𝒋 = ቊ
𝑻𝟏
𝟎 𝒊𝒇 𝒊 ≠ 𝒋
Interpretation:
Multiply two different signals and integrate, the result is 0
Multiply two identical signals and integrate, the result is 1
1) Two orthogonal signals are transmitted together to reduce noise (e.g.
QPSK, QAM)
2) The detector (receiver) is designed based on orthonormality
EE4900/EE6420 Digital Communications Suketu Naik
 21
Energy
 Recall that baseband signal s(t) can be created from K
An example
orthonormal functions as,
𝑲−𝟏
Analog signal
representing 𝒔(𝒕) = ෍ 𝒂𝒌 ∅𝒌 (𝒕)
the bit stream 𝒊=𝟎
Interpretation:
Constants ak are the components of K-dimensional vector so that s(t)
An example
can be represented as K-dim vector S, 2
[𝒂𝟎 𝒂𝟏 𝒂𝟐 ]
𝑺 = [𝒂𝟎 𝒂𝟏 𝒂𝟐 … 𝒂𝑲−𝟏 ]
0
 Energy
𝑲−𝟏 1
𝑻𝟐
𝑬=න 𝒔𝟐 𝒕 𝒅𝒕 = ෍ 𝒂𝟐𝒌
𝑻𝟏 𝒌=𝟎

1) Energy in the signal s(t) is equal to the square of the Euclidean

distance between the K-dimensional point and the origin
2) This result is important for designing the detector (receiver)
EE4900/EE6420 Digital Communications Suketu Naik
 22
EE4900/EE6420: Digital Communications

Modulation
And Detection

EE4900/EE6420 Digital Communications Suketu Naik

 23
Approximation of Received Signal r(t)
 Transmitted signal=s(t), Received signal=r(t), Noise=w(t)
r(t)=s(t) + w(t)
ො (t) to r(t) using basis functions
 Goal: find the best approximation 𝒓
Why?
Because s(t) is corrupted by noise and we need to find an equivalent
of the corrupted s(t) or r(t) in Span{B}
Why?
-Span{B} contains the basis functions
-So that we can determine the symbols and hence the basis functions
Why?
Because the basis functions were used to create symbols which
represented the bits (information)
s(t) r(t)

Symbols representing bits Symbols corrupted by noise

EE4900/EE6420 Digital Communications Suketu Naik
 24
Least-Mean-Square Error
 Transmitted signal=s(t), Received signal=r(t), Noise=w(t)
r(t)=s(t) + w(t)
ො (t) to r(t) using basis functions
 Goal: find the best approximation 𝒓
How?
Find the energy of the error between the received signal r(t) and
approximation 𝒓ො (t) 𝑲−𝟏

𝒔(𝒕) = ෍ 𝒂𝒌 ∅𝒌 (𝒕) Transmitted Vector

𝒌=𝟎 [𝒂𝟎 𝒂𝟏 𝒂𝟐 … 𝒂𝑲−𝟏 ]
𝑲−𝟏

𝒓ො (𝒕) = ෍ 𝒙𝒌 ∅𝒌 (𝒕) Approximate Vector

𝒌=𝟎 [𝒙𝟎 𝒙𝟏 𝒙𝟐 … 𝒙𝑲−𝟏 ]
𝑻𝟐
Energy of the error 𝜺 = න |𝒓 𝒕 − 𝒓ො (𝒕)|𝟐 𝒅𝒕
𝑻𝟏

Energy of the error is minimized by selecting the

approximate vector [𝒙𝟎 𝒙𝟏 𝒙𝟐 … 𝒙𝑲−𝟏 ]
EE4900/EE6420 Digital Communications Suketu Naik
 25
Signal Space Projections or Phase Diagram or Constellation
 Goal: find the approximate vector xk
𝑲−𝟏 𝑻𝟐
Synthesis Analysis
Equation 𝒔(𝒕) = ෍ 𝒂𝒌 ∅𝒌 (𝒕) Equation
𝒙𝒌 = න 𝒓 𝒕 ∅𝒌 𝒕 𝒅𝒕
𝑻𝟏
𝒌=𝟎

An example with K=M=2

Received Signal
Things to remember:
1) Phase diagram shows Basis Function 2
Points
2) Points are denoted by
Vectors from origin
3) Points and Vectors
represent an actual analog
Approximate
baseband signal Vector, xk
4) The amplitude of the
Approximate
baseband signal = point in
Basis Function 1 Received Signal
the phase diagram

EE4900/EE6420 Digital Communications Suketu Naik

 26
Modulator based on Synthesis Equation
Same group of bits
Constructing one
go to each path
symbol so(t)

Serial to Parallel Pulse-shaping filter

can be used here

Data/Audio/Video
signal s(t) is created
[101 100 001 000 010 011 110 111]
from K-symbols
Group of
3 bits Example

One look-up table

per symbol

1) Group of bits (e.g. 101)= the

decimal index (e.g. 5)
2) Now the coefficient with that
index is selected

EE4900/EE6420 Digital Communications Suketu Naik

 27
Summary: Modulation
It’s simple!
 M-ary Modulation:
1) Serial bit stream represents data, audio, video
2) Create group of bits (Serial to Parallel) according to the M
3) Use the group of bits as the index in a look-up-table of
coefficients (amplitudes) of the basis functions
4) Pass the coefficient samples through pulse-shaping filter (one of
the pulse shapes, next lecture). Pulse shape is either the basis
function or a part of basis function. The output is called a
symbol.
5) Add all the symbols together (number of symbols depends on M
which depends on group of bits). This is the baseband signal.
6) Baseband signal then can modulate the carrier signal.
7) Carrier signal is transmitted…

EE4900/EE6420 Digital Communications Suketu Naik

 28
Detector based on Analysis Equation
Find the Subtract the transmitted symbols
Find the approximate from the approximate vector and
area vector find the minimum
Decision
Basis
Rule (derived from
Function
Least Mean Square
Error = the Energy of
the Error)
𝑻𝟐
Received
𝜺 = න |𝒓 𝒕 − 𝒓ො (𝒕)|𝟐 𝒅𝒕
Signal 𝑻𝟏

EE4900/EE6420 Digital Communications Suketu Naik

 29
Practical Detector: Matched Filter
 Matched filter (matched to the pulse-shaping filter) can be used
for detection
Direct Application of Matched Filter
Analysis Equation

EE4900/EE6420 Digital Communications Suketu Naik

 30
Comparison of Two Approaches
 Impulse response of the matched filter=time-reversal of pulse-shaping filter*

Pulse or the impulse

response of the pulse-
shaping filter*

*= this is generally true. For PAM, pulse shape=basis function. However, note that BPSK, QPSK, 16-QAM may not have identical pulse shape
and basis function. For example for QPSK, basis functions=p(t)*cos and -p(t)*sin; where p(t) is the pulse or the impulse response of the pulse-
shaping filter. We will cover pulse shapes, PAM, QAM in upcoming lectures.
EE4900/EE6420 Digital Communications Suketu Naik
 31
Summary: Demodulation (Detection)
It’s simple!
 M-ary Demodulation (Detection):
1) Carrier signal is filtered, down-converted, filtered to baseband
signal
2) Now we reverse the process of modulation: pass it through
matched filter whose impulse response is the time-reversal of the
pulse shaping filter.
3) Output samples (x) from the matched filter are passed through
decision block which implements Least-Mean-Square error logic
4) Least-Mean-Square error logic:
1) x = we hope that these are the symbol amplitudes
2) a = we transmitted these symbol amplitudes=constellation
points
3) Take the differences x-a (Euclidean distance) and find the
minimum. This is the output or our best guess for what was
transmitted.
EE4900/EE6420 Digital Communications Suketu Naik
 32

Examples

EE4900/EE6420 Digital Communications Suketu Naik

 33
Example 5.1.1: Synthesis (Modulator)
K=2=two symbols
basis function 2
Two Orthonormal basis function 1
(note: it can have different
Basis Functions amplitudes than 1 and -1)

Index Coeff

Coefficient chosen according

to the group of bits(=index)

0.5 1 0.5 1 0.5 1

Symbols are constructed

from sum of the two basis
Signal Space Projections functions above
or Phase Diagram or
Constellation
Note that the points So and S1
are actual signals (symbols)
in time
Coefficient chosen according
to the group of bits(=index) 0.5 1 0.5 1 0.5 1

EE4900/EE6420 Digital Communications Suketu Naik

 34
Example 5.1.1: Analysis (Detector)
How do we figure out what was transmitted given r(t)? Received Signal r(t)
A: Find approximate vector [x0 x1]
Analysis Equation
𝑻𝟐
𝒙𝒌 = න 𝒓 𝒕 ∅𝒌 𝒕 𝒅𝒕
𝑻𝟏

Multiply r(t) with basis function,

then find the area (integral)

𝑲−𝟏 The approximate vector

𝒓ො (𝒕) = ෍ 𝒙𝒌 ∅𝒌 (𝒕) shown on the phase diagram
𝒌=𝟎 𝟐
𝒓ො 𝒕 = 𝟐∅𝟎 𝒕 + ∅𝟏 (𝒕)
𝟑

EE4900/EE6420 Digital Communications Suketu Naik

 35
Example 5.1.1: Analysis (Detector)
Transmitted signal s(t) s(t)
6
Received signal r(t)
Shape is changed and
4
Amplitude is changed. Why?

2 2
2

+ =
0.5 1 0.5 1 0.5 1

Approximate signal 𝒓ො (𝒕) 8/3=2.67

4/3=1.33
1 1

0.5 1 0.5 1 0.5 1

-1

EE4900/EE6420 Digital Communications Suketu Naik

 36
Example 5.1.1: Analysis (Detector)
Approximate vector
x=[2 2/3]
s(t)
What was transmitted? 6
Sm=[6 2]
2

0.5 1

What did we determine as the transmitted signal?

Decision Rule:

|2-6|=4
|2/3-2|=1.33
Answer: Minimum of {4, 1.33}=1.33
What was transmitted? 0d (000b) or 1d (001b)or 2d (010b)?

EE4900/EE6420 Digital Communications Suketu Naik