Pulse Modulation Chapter 5

Pulse Modulation:
Professor Deepa Kundur
Transition from Analog to Digital
University of Toronto
Communications

Professor Deepa Kundur (University of Toronto) Pulse Modulation 1 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 2 / 61

Section 5.1

Pulse Modulation Sampling

Reference:
From duality:
Sections 5.1- 5.6 of
periodic in time discrete in frequency
S. Haykin and M. Moher, Introduction to Analog & Digital discrete in time periodic in frequency
Communications, 2nd ed., John Wiley & Sons, Inc., 2007. ISBN-13
978-0-471-43222-7.

Professor Deepa Kundur (University of Toronto) Pulse Modulation 3 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 4 / 61
 Section 5.1 Section 5.1

Sampling Sampling
We can model sampling as multiplication of an analog waveform g (t)
with an impulse train:
I analog waveform: g (t)
I impulse train: ∞
∞ X
X gδ (t) = g (t) · δ(t − nTS )
δ(t − nTS )
n=−∞
n=−∞
∞
X
I model of sampling: = g (nTS )δ(t − nTS )
n=−∞
∞
X
gδ (t) = g (t) · δ(t − nTS ) Note: gδ (t) contains the information of g (nTS ) and represents a
n=−∞ good model of sampling.
∞
X
= g (nTS )δ(t − nTS )
n=−∞

Professor Deepa Kundur (University of Toronto) Pulse Modulation 5 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 6 / 61

Section 5.1 Section 5.1

g(t)

t F
gδ (t) ←→ Gδ (f )
F
g (t) ←→ G (f )
S
impulse train
∞ ∞
X F 1 X k
1 δ(t − nTS ) ←→ δ(f − )
n=−∞
TS TS
k=−∞
t
∞ ∞
-5TS -4TS -3TS -2TS -TS 0 TS 2TS 3TS 4TS 5TS X F 1 X k
gδ (t) = g (t) · δ(t − nTS ) ←→ Gδ (f ) = G (f ) ? δ(f − )
S S n=−∞
TS TS
g(0)
k=−∞
g(-TS) g(TS) impulse train ∞ ∞
g(2TS) scaled by signal X F 1 X k
samples gδ (t) = g (t)δ(t − nTS ) ←→ Gδ (f ) = G (f ) ? δ(f − )
-5T -4T
S S 4TS 5TS
t n=−∞
TS TS
k=−∞
-3TS -2TS -TS 0 TS 2TS 3TS
∞ ∞
X F 1 X k
gδ (t) = g (nTS )δ(t − nTS ) ←→ Gδ (f ) = G (f − )
n=−∞
TS TS
k=−∞
g(0)
g(-TS) g(TS)
g(2TS) signal samples

-5 -4 -3 3 4 5
k
Professor Deepa Kundur (University of Toronto) -2 -1 Modulation
Pulse 0 1 2 7 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 8 / 61
 Section 5.1 Section 5.1

1
g(t) Let fs = TS :
∞ ∞
X F 1 X k
gδ (t) = g (nTS )δ(t − nTS ) ←→ Gδ (f ) = G (f − )
t
n=−∞
TS TS
k=−∞
∞ ∞
F
X X
G(f )
gδ (t) = g (nTS )δ(t − nTS ) ←→ Gδ (f ) = fS G (f − kfS )
n=−∞ k=−∞
A
F
f
Recall, δ(t − t0 ) ←→ e j2πt0 f
∞
" #
X
g (t) Gδ (f ) = F [gδ (t)] = F g (nTS )δ(t − nTS )
g(0)
g(-TS) g(TS) n=−∞
g(2TS) ∞ ∞
X X
-5T -4T
S S 4TS 5TS
t = g (nTS )F [δ(t − nTS )] = g (nTS )e j2πnTS f
-3TS -2TS -TS 0 TS 2TS 3TS
n=−∞ n=−∞

Therefore,
|G (f )|
∞
X ∞
X
A/Ts Gδ (f ) = g (nTS )e j2πnTS f = fS G (f − kfS )
f
n=−∞ k=−∞
-1/Ts 0 1/Ts
Professor Deepa Kundur (University of Toronto) Pulse Modulation 9 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 10 / 61

Section 5.1 Section 5.1

|G (f )| |G (f )|

A/TS A/TS

f f
-1/TS 0 1/T S -1/TS 0 1/T S
Increasing Ts

Increasing Ts
|G (f )| |G (f )|

A/TS A/TS

f f
-2/TS -1/T 0 1/T S 2/TS -2/TS -1/T 0 1/T S 2/TS
S S

2A/TS |G (f )| ALIASING 2A/TS |G (f )| ALIASING

A/TS A/TS

f f
-4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS -4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS
S S

Professor Deepa Kundur (University of Toronto) Pulse Modulation 11 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 12 / 61
 Section 5.1 Section 5.1

|G (f )| |G (f )|

A/TS A/TS

f f
-1/TS 0 1/T S -1/TS 0 1/T S
Increasing Ts

Increasing Ts
|G (f )| |G (f )|

A/TS A/TS

f f
-2/TS -1/T 0 1/T S 2/TS -2/TS -1/T 0 1/T S 2/TS
S S

2A/TS |G (f )| ALIASING 2A/TS |G (f )| ALIASING

A/TS A/TS

f f
-4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS -4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS
S S

Professor Deepa Kundur (University of Toronto) Pulse Modulation 13 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 14 / 61

Section 5.1 Section 5.1

|G (f )| |G (f )|
ANTI-ALASING
A/TS A/TS FILTER APPLIED
f f
-1/TS 0 1/T S -1/TS 0 1/T S
Increasing Ts

Increasing Ts
|G (f )| |G (f )|

A/TS A/TS

f f
-2/TS -1/T 0 1/T S 2/TS -2/TS -1/T 0 1/T S 2/TS
S S

2A/TS |G (f )| ALIASING |G (f )|

A/TS A/Ts

f f
-4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS -4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS
S S

Professor Deepa Kundur (University of Toronto) Pulse Modulation 15 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 16 / 61
 Section 5.1 Section 5.1

|G (f )| |G (f )|

A/TS A/TS

f f
-1/TS 0 1/T S -1/TS 0 1/T S
Increasing Ts

Increasing Ts
|G (f )| |G (f )|

A/TS A/TS

f f
-2/TS -1/T 0 1/T S 2/TS -2/TS -1/T 0 1/T S 2/TS
S S

2A/TS |G (f )| ALIASING 2A/TS |G (f )| ALIASING

A/TS A/TS

f f
-4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS -4/T -3/TS -2/TS -1/TS 0 1/T S 2/T S 3/T S 4/TS
S S

Professor Deepa Kundur (University of Toronto) Pulse Modulation 17 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 18 / 61

Section 5.1 Section 5.1

|G (f )|
Sampling Theorem
A/TS
1
W < −W
TS
f W < fS − W
-W 0 W 1/T S-W 1/T S 1/T S+W fS > 2W = Nyquist Rate
Increasing Ts

|G (f )| OR

ALIASING A/TS fS > 2W = Nyquist Rate

1
TS < = Maximum Sampling Period
2W
f
-W 0 W 1/T S
Professor Deepa Kundur (University of Toronto) 1/T
Pulse-W
S Modulation
1/T S+W 19 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 20 / 61
 Section 5.1 Section 5.1

Sampling Theorem
Suppose that a signal g (t) is strictly band-limited with no frequency components
higher than W Hz. That is, G (f ) is zero for |f | ≥ W .

Then g (t) can be exactly recovered from its sample values g (nTS ) for
n = 0, ±1, ±2, ±3, . . . through band-limited interpolation if:
Pulse Modulation
1
fS = > 2W
TS
where 2W is called the Nyquist Rate.

We will assume that all message signals m(t) from now on are sampled above the
Nquist Rate.

Professor Deepa Kundur (University of Toronto) Pulse Modulation 21 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 22 / 61

5.2 5.2

Pulse Modulation Pulse Amplitude Modulation (PAM)

I the variation of a regularly spaced constant amplitude pulse
stream to superimpose information contained in a message A
t
signal. T t
T
Ts
Ts Note: T < Ts
m(-Ts ) m(0)
A m(t)
m(t)
m(-Ts ) m(0)
m(Ts )
m(Ts ) m(2Ts )
m(2Ts )
t
T t
t
Ts Note: T < Ts

I Three types: m(-Ts ) m(0) m(Ts )

m(t)
m(-Ts ) m(0) m(Ts ) m(2Ts )
1. pulse amplitude modulation (PAM) m(t)
m(2Ts )
2. pulse duration modulation (PDM) s(t)
s(t) t
3. pulse position modulation (PPM) t
T
T
Professor Deepa Kundur (University of Toronto) Pulse Modulation 23 / 61 Professor Deepa Kundur (University T
ofs Toronto) Pulse Modulation 24 / 61
Ts
 5.2 5.2

Pulse Amplitude Modulation Sample-and-Hold Analysis

t Suppose
T 1
Two steps: Sampling and
Ts Hold
!
t − T2

1 0<t<T
h(t) = rect =
m(-Ts ) m(0)
1. Instantaneous sampling: the message
m(Ts ) m(t) is sampled every Ts seconds T 0 otherwise
m(t)
1 m(2Ts )
where fs = Ts obeys the sampling theorem. T

2. Lengthening: extending the duration of each sample so that it ∞

t occupies T X
seconds. mδ (t) = m(nTs )δ(t − nTS )
n=−∞
∞
" #
m(-Ts ) m(0) m(Ts ) X
m(t) mδ (t) ? h(t) = m(nTs )δ(t − nTS ) ? h(t)
m(2Ts )
n=−∞
s(t) ∞
t X
T
= m(nTs )δ(t − nTS )?h(t)
n=−∞
Ts ∞
X
= m(nTs )h(t − nTS )
n=−∞

Professor Deepa Kundur (University of Toronto) Pulse Modulation 25 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 26 / 61

5.2 5.2

Sample-and-Hold Analysis Sample-and-Hold Analysis

∞
X
s(t) = mδ (t) ? h(t) = m(nTs )h(t − nTS )
n=−∞

t
s(t) = mδ (t) ? h(t)
T mδ (t)
S(f ) = Mδ (f ) · H(f )
Tsm(-T ) m(0) m(T )
m(t) s s F
m(2Ts ) h(t) ←→ H(f )
m(-Ts ) m(0) m(Ts )
m(t)
m(2Ts ) F
t rectangle ←→ sinc
t

s(t) = mδ (t) ? h(t)

m(-Ts ) m(0) m(Ts )

m(2Ts )
s(t)

t
T
Ts

Professor Deepa Kundur (University of Toronto) Pulse Modulation 27 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 28 / 61
 5.2 5.2

Step 1: Instantaneous sampling. Step 2: Lengthening.

m(t) m (t)
m(0) m(TS)
m(-TS)
m(2TS)

4TS 5TS S S 4TS 5TS

t t
-5T -4T -3TS -2TS -TS 0 TS 2TS 3TS -5T -4T -3TS -2TS -TS 0 TS 2TS 3TS

|M(f )| |M (f )|

A/Ts A/Ts

f f
-1/Ts 0 1/Ts -1/Ts 0 1/Ts

m (t) s(t)
m(0) m(TS) m(0) m(TS)
m(-TS) m(-TS)
m(2TS) m(2TS)

4TS 5TS 4TS 5TS

t t
-5TS -4TS -3TS -2TS -TS 0 TS 2TS 3TS -5TS -4TS -3TS -2TS -TS 0 TS 2TS 3TS
T

|M (f )| |S(f )|

scaled
A/Ts A/Ts
by a sinc
Professor Deepa Kundur (University of Toronto) Pulse Modulation f 29 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation f 30 / 61
-1/Ts 0 1/Ts -1/T -1/Ts 0 1/Ts 1/T

5.2 5.2

Recovery.
Pulse Amplitude Modulation |M(f )|

Recovery: f
-1/Ts 0 1/Ts
I Pass the samples s(t) through a lowpass filter.
|M (f )|

A/Ts
There is a trade-off to the pulse width T :
f
I The signal lengthening stage reduces the bandwidth of the -1/Ts 0 1/Ts

overall pulse making it more efficient for communications. |S(f )|

scaled Recovery filter
I However, this is some distortion when recovering the signal as by a sinc Ts
A/Ts

the sinc function in the frequency domain warps the frequency f

-1/Ts 0 1/Ts
domain of the information signal.
|M(f )|
warping due
to sinc function
A

Professor Deepa Kundur (University of Toronto) Pulse Modulation 31 / 61 Professor Deepa Kundur (University-1/Ts
of Toronto) Pulse Modulation
0 1/Ts f 32 / 61
 5.3 5.3

Pulse Duration Modulation (PDM) Pulse Duration Modulation

t I The width of the pulse reflects the sampled signal amplitude.

T
Ts
I the position of the leading edge, trailing edge or both may be
modified to reflect the changing duration of the pulse
m(-Ts ) m(0) m(Ts ) I also known as: pulse width modulation or pulse length
m(t)
m(2Ts )
modulation
t

PDM I PDM is wasteful of energy when the pulses are long, but the
m(-Ts ) m(0) m(Ts ) s(t)
information is only in the pulse transitions
m(t)
m(2Ts )

Professor Deepa Kundur (University of Toronto) Pulse Modulation 33 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 34 / 61

5.3 5.3

Pulse Position Modulation

t
T
Ts
I The position of the pulse reflects the sampled signal amplitude.
m(-Ts ) m(0) m(Ts )
m(t)
m(2Ts ) I PPM can be represented as:
t ∞
X
s(t) = g (t − nTs − kp m(nTs ))
PDM n=−∞
m(-Ts ) m(0) m(Ts ) s(t)
m(t)
m(2Ts )
where
t
I kp is the sensitivity factor
I the adjacent pulses must be strictly non-overlapping
PPM
m(t) s(t)

Professor Deepa Kundur (University of Toronto) Pulse Modulation 35 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 36 / 61
 5.3 5.3

PDM
m(t)
m(-Ts ) m(0) s(t)
Pulse Position Modulation
m(2Ts )

m(Ts )

PPM ∞
X
m(t) s(t)
s(t) = g (t − nTs − kp m(nTs ))
n=−∞
t
where
m(Ts )
kp |m(t)|max < (Ts /2)
T T for strictly non-overlapping pulses.
n=0 n=1
SHIFT FROM SHIFT FROM
HIGHEST LOWEST
MESSAGE MESSAGE
AMPLITUDE AMPLITUDE
t

Ts
Professor Deepa Kundur (University of Toronto) Pulse Modulation
Ts 37 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 38 / 61

5.3 5.4

Pulse Position Modulation Analog vs. Digital Communications

I Channel noise and signal distortion on analog communication
system is cumulative.
Recovery:
1. Determine pulse centers to decode values of {m(nTs )}. I Regenerative repeaters in digital communication system can
2. Use bandlimited interpolation to obtain m̂(t). practical eliminate degrading effects of channel noise and signal
distortion.

I Coding can be used in digital communication systems for greater

Assuming that the samples {m(nTs )} obey the Sampling Theorem, reliability and security.
then m̂(t) = m(t) leading to ideal communication reconstruction.
I Digital communications requires that we not just sample in time,
but quantize in amplitude.

Professor Deepa Kundur (University of Toronto) Pulse Modulation 39 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 40 / 61
 5.5 5.5

Amplitude Quantization Amplitude Quantization

Let:
Amplitude quantization: the process of transforming the sample I m = original discrete-time signal sample
amplitude m(nTs ) of a baseband signal m(t) at time t = nTs into a
I v = quantized digital signal sample
discrete amplitude v (nTs ) taken from a finite set of possible levels.
I g (·) = quantization operator
I non-reversible process
I em = quantization error

I Let us denote m(nTs ) simply as m and v (nTs ) simply as v . v = g (m)

em = v − m

Professor Deepa Kundur (University of Toronto) Pulse Modulation 41 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 42 / 61

5.5 5.5

Uniform Quantization Example Uniform Quantization Example

m(nTs) v(nTs) m(nTs)

1 1

-3 -2 -1 0 1 2 3
n -3 -2 -1 0 1 2 3
n

. . . result of rounding to the nearest quantization level.

Professor Deepa Kundur (University of Toronto) Pulse Modulation 43 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 44 / 61
 5.5 5.5

Uniform Quantization: Midtread Uniform Quantization: Midrise

v v
midtread midrise
3 3
2 2

-3 -2 - 2 3 m -3 -2 - 2 3 m
- -
-2 -2
-3 -3
em quantization error em quantization error

.5 .5

-2.5 -1.5 -.5 .5 1.5 2.5 3.5

m -2.5 -1.5 -.5 .5 1.5 2.5 3.5
m
-.5 -.5
Professor Deepa Kundur (University of Toronto) Pulse Modulation 45 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 46 / 61

5.5 5.6

Uniform Quantization: Midtread vs. Midrise Pulse-Code Modulation

I Most basic form of digital pulse modulation

v v
midtread midrise
3 3
2 2 I Elements of pulse-code modulation (PCM):
-3 -2 - 2 3 m -3 -2 - 2 3 m 1. Transmitter
- -
-2 -2
2. Transmission Path
-3 -3 3. Receiver
em quantization error em quantization error
PCM Data Channel
.5 .5 Sequence Output
m m

DESTINATION
-2.5 -1.5 -.5 .5 1.5 2.5 3.5 -2.5 -1.5 -.5 .5 1.5 2.5 3.5
SOURCE
Tranmission
-.5 -.5 Transmitter Receiver
Path

Professor Deepa Kundur (University of Toronto) Pulse Modulation 47 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 48 / 61
 5.6 5.6

PCM Transmitter PCM Transmitter: Sampler

Continuous-time Anti-aliased Continuous-time Anti-aliased
Message Cts-time Discrete-time Digital PCM Data Message Cts-time Discrete-time Digital PCM Data
Signal Signal Signal signal Sequence Signal Signal Signal signal Sequence
Low-pass Low-pass
Source Sampler Quantizer Encoder Source Sampler Quantizer Encoder
Filter Filter

{
{
{
{

{
{
{
{
Anti-aliasing Sampling above Using a Maps Numbers Anti-aliasing Sampling above Using a Maps Numbers
Filter Nyquist with Non-uniform to Bit Sequences Filter Nyquist with Non-uniform to Bit Sequences
Narrow Rectangular Quantizer Narrow Rectangular Quantizer
PAM Pulses PAM Pulses

m(-Ts ) m(0) m(Ts ) m(-Ts ) m(0) m(Ts )

m(t)
m(2Ts ) m(2Ts )

s(t)
t t
T
Ts

Professor Deepa Kundur (University of Toronto) Pulse Modulation 49 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 50 / 61

5.6 5.6

PCM Transmitter: Non-Uniform Quantizer PCM Transmitter: Non-Uniform Quantizer

Continuous-time Anti-aliased
Message Cts-time Discrete-time Digital PCM Data
Signal Signal Signal signal Sequence
Compressor: µ-law:
Low-pass
Source Sampler Quantizer Encoder
Filter
{
{
{
{
Anti-aliasing Sampling above Using a Maps Numbers
Filter Nyquist with Non-uniform to Bit Sequences m ≡ message sample
Narrow Rectangular Quantizer
PAM Pulses v ≡ quantized value
log ≡ natural logarithm
Amplitude Uniform log(1 + µ|m|)
Compressor Quantizer |v | =
v log(1 + µ)
3
2
1.0
-3 -2 - 2 3 m
-
0.75
large mu
-2
-3 Note: µ = 0 corresponds to a linear quantizer.
Normalized output |v|

0.5
v[n] m[n] Typically µ ≈ 255 used in practice.
1
0.25

-3 -2 -1 0 1 2 3
n
0 0.25 0.5 0.75 1.0
Normalized input |m|

Professor Deepa Kundur (University of Toronto) Pulse Modulation 51 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 52 / 61
 5.6 5.6

PCM Transmitter: Non-Uniform Quantizer PCM Transmitter: Non-Uniform Quantizer

Compressor: µ-law:
1.0 1.0 Compressor: A-law:

inc
0.75 rea 0.75 
sin large mu A|m| 1
Normalized output |v|

Normalized output |v|

g mu 
 1+log(A)
0 ≤ |m| ≤ A

0.5 0.5
|v | =
1+log(A|m|) 1
≤ |m| ≤ 1


mu=0 1+log(A) A
0.25 0.25

Note: A = 1 corresponds to a linear quantizer.

0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0
Typically A ≈ 100 used in practice.
Normalized input |m| Normalized input |m|

Professor Deepa Kundur (University of Toronto) Pulse Modulation 53 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 54 / 61

5.6 5.6

PCM Transmitter: Non-Uniform Quantizer PCM Transmitter: Encoder

Continuous-time Anti-aliased
Message Cts-time Discrete-time Digital PCM Data
Signal Signal Signal signal Sequence
Compressor: µ-law: Low-pass
1.0 1.0 Source Sampler Quantizer Encoder
Filter

{
{
{
{
inc Anti-aliasing Sampling above Using a Maps Numbers
rea large A
0.75 sin 0.75 Filter Nyquist with Non-uniform to Bit Sequences
Normalized output |v|

Normalized output |v|

g A Narrow Rectangular Quantizer

PAM Pulses
0.5 0.5
A=1
I maps quantization-level output to a code word
0.25 0.25
I typically binary code words are employed

0 0.25 0.5 0.75 1.0 0 0.25 0.5 0.75 1.0

Normalized input |m| Normalized input |m|

Professor Deepa Kundur (University of Toronto) Pulse Modulation 55 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 56 / 61
 5.6 5.6

Encoder: Example PCM: Transmission Path

PCM Data Channel
8 Quantization-levels, or R = 3-bit code words: Sequence Output

DESTINATION
Tranmission

SOURCE
Transmitter Receiver
Path
Quantization-Level Index Binary Codeword

0
1
000
001
PCM Data
Shaped for
Transmission
Tranmission
Line
Regenerative
Repeater
Tranmission
Line
... Regenerative
Repeater
Tranmission
Line
Channel
Output
2 010
3 011
4 100
5 101
6 110 Distorted
Amplifier- Decision-making
Regenerated
PCM PCM
7 111 Wave Equalizer Device Wave

Timing
Circuit

Professor Deepa Kundur (University of Toronto) Pulse Modulation 57 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 58 / 61

5.6 5.6

PCM: Regenerative Repeater PCM: Regenerative Repeater

Distorted Regenerated
Amplifier- Decision-making
PCM PCM
Wave Equalizer Device Wave
I Amplifier-Equalizer: shapes the received pulses to compensate
for amplitude and phase distortions produced by transmission line Timing
Circuit
I Timing Circuit: produces periodic pulse train derived from Original PCM Wave
received pulses to sample the pulses where signal-to-noise ratio t
is highest
t
I Decision-making Device: sample of pulse is compared to a
pre-determined threshold t
I if threshold exceed a clean new pulse representing 1 is
transmitted THRESHOLD t
I otherwise a clean new pulse representing 0 is transmitted “1” “0” “0” “1” “1” “0”
t

BIT ERROR
Professor Deepa Kundur (University of Toronto) Pulse Modulation 59 / 61 Professor Deepa Kundur (University of Toronto) Pulse Modulation 60 / 61
 5.6

PCM: Receiver
Two Stages:

1. Decoding and Expanding:

1.1 regenerate the pulse one last time and interpret bit sequence
1.2 group bits into code words
1.3 interpret code words as quantization level
1.4 pass level through expander (opposite of compressor)

2. Reconstruction:
2.1 pass expander output through low-pass reconstruction filter
(cutoff is equal to message bandwidth) to estimate original
message m(t)


Professor Deepa Kundur (University of Toronto) Pulse Modulation 61 / 61