Signal Transmission &

Reception Techniques

Chapter 1
An introduction to communication systems

1.2: Overview of a PAM digital system

Suggested Reading: Telecommunication Breakdown: Ch. 1 - 2

What we will see in this course… and what we won’t
 What we will see:
 Introduction to the different methods for information exchange in digital format at
the Physical Layer (PHY) level: how to map bits to analog signals and vice versa
 Focus on Pulse Amplitude Modulation (PAM) as illustrative example
 Overview of main digital TX / RX elements, channel induced phenomena and
performance metrics

TCP/UDP Transport

IP Network Analog
Baseband Front
Processing End
PPP Link

Medium Access
Analog
Application signal
bits Packets
Higher Physical Layer “Transmission
Layers (PHY) medium”

 What will not be covered:

 Other more advanced modulations (multicarrier, CDMA, MIMO systems, etc.)
 We won’t go into details about source and channel coding
 Higher layers will not be covered (but they are also important)
Contents

1. Overview of a digital PAM transmitter

 Channel coding
 Line coding: from bits to symbols
 Transmission pulses
 Analog front end

2. Overview of a digital PAM receiver

 Tasks of a digital PAM receiver
 Elements of a communication system

Analog
message bits bits Channel bits
Source Digital
A/D encoder
encoder transmitter
(rate k/n)
Compression: Protection: Analog
eliminates redundancy Introduces redundancy signal
in a controlled manner

Physical channels accept and delivers analog signals only: Channel

Every digital comunication system
has an analog core.
Analog
signal

Reconstructed bits bits bits

Source Channel
analog D/A Receiver
decoder decoder
message
Effect of the encoder on the bit rate

 Let Rb (bits/s) be the bit rate at the input of the channel encoder
 For each input block of length k bits, the encoder delivers a block of n
bits at its output (n > k)
 The code rate is therefore k/n < 1

 Bit rate at the encoder output: R’b (bits/s)

sistematic bits parity bits

1 2 ··· k
Channel 1 2 ··· k k+1 ··· n
encoder

Time duration = k /Rb (s) Time duration = n /R’b (s)

 The time duration of input and output blocks must be the same
(otherwise buffers would overflow):
Digital PAM (Pulse Amplitude Modulation) Transmitter

Clock signal with

clock frequency Rs

Analog
symbols baseband
bits bits Line PAM signal
Channel Transmit
encoder Encoder
(Bits → Symb)
pulse g(t)

rate Rb (bits/s) R’b (bits/s) Rs (símb/s)

Bandpass
Mixer Amplif.
Filter (fc)
RF Signal

Local

~ oscillator
(fc)
RF stage
(analog front end)
Mapping bits to symbols: line codes

 Bits at the channel encoder output are grouped in blocks of b bits

 A symbol (voltage level) is assigned to each possible b-bit block

 Constellation: the set of all possible symbols. Its size is M = 2b

 Therefore each symbol carries b = log2M information bits

M = 4 (4-PAM)
 Examples of line codes: Bits Symbol (V)
M = 2 (2-PAM or BPSK)
00 -3a
 M-PAM: the constellation is Bit Symbol (V)
01 -a
{ ± a, ± 3a, ···, ± (M-1)a } 0 -a
10 +a
1 +a 11 +3a

PAM : Pulse Amplitude Modulation

BPSK: Binary Phase-Shift Keying
Bit rate vs Symbol rate

bits bits symbols

Channel Line
encoder encoder
(Bits → Symb)

rate Rb (bits/s) R’b (bits/s) Rs (símb/s) Émile Baudot

(1845-1903)

 The time duration of a symbol, T, is known as the symbol interval

(or baud interval)

 Its inverse is the symbol rate (or baud rate): Rs = 1/T (symb/s or
bauds)

 Therefore R’b and Rs are related as

R’b (bits/s) = Rs (symb/s) · log2M (bits/symb)

Example: from ASCII to 4-PAM
 How can we assign 4-PAM symbols to alphanumeric characters?

 Standard ASCII: 1 character = 8 bits (this is a kind of source encoding)

 Assume no channel encoding

 If 4-PAM is used, then M = 4 and b = log2M = 2 bits/symbol

 For each alphanumeric character, we need to transmit 4 symbols

Bits Symbol (V) Character ASCII symbols

00 -3a a 01 10 00 01 -a +a -3a -a
01 -a b 01 10 00 10 -a +a -3a +a
10 +a c 01 10 00 11 -a +a -3a +3a
11 +3a ··· ··· ···

 With 8-PAM (3 bits/símb) ⇒ no integer no. of PAM symbols per character…

 Not a problem: the bit stream generated by the ASCII encoder is chopped into
bit triplets, so that to each triplet an 8-PAM symbol is assigned
Building the baseband PAM signal

 The symbols to transmit, {A[n]}, constitute a discrete sequence...

however, the channel only handles analog signals

 The symbol sequence {A[n]} is mapped into a sequence of

analog pulses, based on a basic transmission pulse g(t)

 Every T seconds (a symbol time duration), a pulse g(t-nT-t0) is

generated, and then it is scaled by the n-th symbol A[n]

 The analog signals built in this way is the superposition of all these
pulses:

where t0 is arbitrary.
Example: rectangular pulses

 g(t) could be taken as a rectangular pulse with duration T seconds:

4-PAM
BPSK
s(t) s(t) A[1] = 3a A[4] = 3a
3a
A[0] = a A[3] = a
+a A[0] = a
+a
t0+3T

t0 t0+T t0+2T t0+3T t t0 t0+T t0+2T t

-a
A[2] = -a
-a
A[1] = -a A[2] = -a
-3a
A[3] = -3a

 Other types of pulses g(t) are possible (and in fact desirable)

Analog front end: frequency traslation, amplification, & filtering
 Generally the Fourier Transform G(f ) of the basic pulse g(t) is lowpass
 For instance: g(t) = rect(t/T) ↔ G(f ) = T·sinc(Tf )

T |G(f )|
g(t)
1

t f
-T/2 0 T/2 ··· -2/T -1/T 0 1/T 2/T ···

 We’ll see that this implies that s(t) is also lowpass. If the channel excessively attenuates
the low frequency components, it’s not a good idea to directly inject s(t)
 Modulation: generation of a bandpass signal x(t) (modulated signal) from the lowpass
signal s(t) (modulating signal)
 Amplification: Channel attenuation ⇒ the power of x(t) must be amplificada to obtain
an adequate level at the receiver.
Keep in mind that the maximum transmit power is limited!
 Filtering: Once amplified, the bandpass signal may be bandpass filtered to make sure
that we are not transmitting outside our assigned spectral range
 Demodulation: at the receiver, we obtain a more or less distorted replica of x(t), and
from it we must recover s(t) as accurately as possible
Contents

1. Overview of a digital PAM transmitter

 Channel coding
 Line coding: from bits to symbols
 Transmission pulses
 Analog front end

2. Overview of a digital PAM receiver

 Tasks of a digital PAM receiver
 Clock, carrier, frame synchronization
 Gain control
 Filtering
Introduction

bits bits bits

Reconstructed
Received Channel Source
Receiver D/A analog
signal decoder decoder message

 Which functions must the receiver implement to recover the

transmitted bits?
 We’ll see some of them over the course

 How can such functions be implemented?

 Adequate combination between analog and digital signal processing

 In this chapter we will have an overview of the workings of a

simple PAM receiver
Ideal lowpass channel

 An idealized scenario: the channel is lowpass and it only introduces

attenuation 1/c (thus c is the gain) and delay η

 Then no modulation is needed ⇒ the transmitted signal is directly

the baseband PAM signal:

 Received signal (taking into account attenuation and delay):

 How can we recover the original information (bits) from the

received analog signal r(t)?
Baseband PAM signal

···
···
t
T

 It makes sense to first recover the symbols {A[n]}, and from them
decode the bits. For instance, we could:

1. Sample r(t) taking one sample every T seconds

2. Quantize the samples to determine which symbols were transmitted
3. Recover the original bits from these recovered symbols

 This is indeed the basic philosophy… although it is not as simple as

it looks!
 Uncertainty about many parameters of the received signal makes this
task difficult
Clock recovery

 What is the best way to sample r(t) taking one sample per
symbol?

···
··· ta+2T
t
ta ta+T ta+3T

···
··· tb+2T
t
tb tb+T tb+3T
Gain control

 With BPSK (M = 2), we just need to check if the samples are

positive or negative to decide about the original symbol

 But with larger constellations ( M > 2 ), accurately compensating

the attenuation of the channel becomes critical

s(t) r(t) = c· s(t - η)

+3a +3a

+a +a

t t
-a -a η

-3a -3a
Frame synchronization

 Assume transmission of alphanumeric characters using 8-bit ASCII

and a 4-PAM line encoder

 Suppose that after scaling and sampling, the observed sequence

of received symbols is:
···, -a, -a, +a, -3a, -a, ···

 Recall that, with this line code, four symbols comprise one ASCII
character

 Two possible ways to recover the message:

 Should we decide that the original symbol was “X”, or that it was
“a” ?
Carrier recovery

 When the spectral content of the baseband PAM signal must be

shifted to a different frequency band, the TX includes a modulator in
the analog front end (“upconversion”)

 The RX should “downconvert” the received signal to the baseband as

a previous step to sampling and decoding. How can this be done?
 Amplitude Modulation (AM): multiply by sine wave, then lowpass filter
 Frequency or phase modulation (FM, PM): slightly more involved…

 For AM, we’ll see that correct operation requires that the TX and RX
oscillators have
 Exactly the same frequency fc
 And also the same phase φ

 This is impossible in practice, so means must be provided for the RX

to “lock” its oscillator in phase and frequency to that of the TX, based
on the received signal
Filtering

 In practice, channels are not ideal. They introduce degradation:

 Background noise
 Interferences (man-made signals from outside the system)
 Frequency-dependent attenuation (wideband wired and wireless systems)
 Multipath (wireless systems)

 The received signal is a

distorted version of the
transmitted signal

 The RX should try to

compensate using filters
(matched filters,
equalizers…)