S-72.

245 Transmission Methods in

Telecommunication Systems (4 cr)

Sampling and Pulse Coded Modulation

 Sampling and Pulse Coded Modulation
 History
 Encoding types
 Standards
 Modules in PCM-link
 Sampling and reconstruction
– Time and frequency domain
– Idea sampling
– Chopper sampling
– Aliasing
 Quantization
 PCM vrs analog modulation
 Repeaters and line coding
 PMC encoding and decoding
circuits TDM: Time Division Multiplexing
 Digital hierarchies: PDH FDM: Frequency Division Multiplexing
(plesiochronous digital PAM: Pulse Amplitude Modulation
hierarchy) T1/E1 PCM: Pulse Coded Modulation

2 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Short history of pulse coded modulation

 A problem of PSTN analog techniques (eg SSB-FDM) was that

transmitting multiple channels was difficult due to non-linearities
resulting channel cross-talk
 1937 Reeves and Delorane ITT labs. tested TDM-techniques by
using electron-tubes
 1948 PCM tested in Bell Labs: Using this method it is possible
to represent a 4 kHz analog telephone signal as a 64 kbit/s
digital bit stream ( 2 x 4 kHz x 8 bits = 64 kbit/s)
 TDM was taken into use in 1962 with a 24 channel PCM link
 The first 30-channel PCM system installed in Finland 1969
 Nowadays all exchanges in Finland use ISDN & PCM based
cables, microwave or optical links

3 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

PCM coding is a form of waveform coding

Voice coders

 Waveform coders reply signal by quantized (discrete) values,

- precise waveform replay but requires a lot of bandwidth
 Parameterized coders count on system model that
reproduces the signal. Only model parameters are transmitted
and updated. Very low rate can be obtained but this is paid by
quality degradation
 Hybrid coders (as D- modulation) are a compromise solution
4 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 speech/video coding standards
Some important ITU-T

5 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Pulse Coded Modulation (PCM)
 PCM is a method by which analog message can be transformed
into numerical format and then decoded at the receiver
transmitter

LP-filtering sampling quantization PAM/HDB-3

Channel
receiver

LP-filtering PAM/HDB-3 Pulse

regeneration

6 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

7 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
8 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
9 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
10 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
11 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
12 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
13 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 fs  W
Aliasing and sampling theorem
 Nyqvist sampling theorem:
W fs 2 fs

If a signal contains no frequency components for f  W

it is completely described by instantaneous uniformly spaced
time samples having periodTs  1/ 2W . The signal can hence
been reconstructed from its samples by an ideal LPF of
bandwidth B such that W  B  f s  W.

 Note: If the message contains higher frequencies than Wmax=

fs/2 they will also be present at the sampled signal! An
application of this is the sampling oscilloscope

 Also, it follows from the sampling theorem that

Two pieces of independent information / second (independent
samples) can be transmitted in 1 Hz wide channel

because signal having bandwidth B can be constructed from

rate 2B independent samples
14 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
1. Pulses in practical sampling waveform have finite duration and
rise times -> linear distortion
2. Reconstruction filters are not ideal lowpass filters
-> spectral folding
3. Sampled signal spectra is not frequency limited -> spectral
folding
4. Samples digitized by finite length words -> quantization noise

Spectral folding

15 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Quantization  Original signal
has continuous
amplitudes in its
Analog dynamic range
signal to be  PAM - signal is a
transmitted
discrete constant
period, pulse train
having continuous
Continuous amplitude values
PAM-pulse
train  Quantized PAM
signal has only the
values that can be
quantized by the
words available
Quantized
(here by 3 bit
PAM-pulse
train words)

16 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Uniform quantization: transmitter
 Transforming the continuos samples into discrete level samples
is called quantization
 In uniform quantization quantization step size is constant

PCM transmitter
Q-PAM
(quantized signal PAM Q-PAM
amplitude)

M  2, v  8
q  M v  256 (v  log M q )
2 / q  7.8 103
PAM (analog signal q: number of quantization levels
amplitude) v : number of quantization bits
ADC: Analog-to-Digital Converter
r: output bitrate

17 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Reconstruction from the quantized signal

PCM receiver
Q-PAM  Note that quantization
 q  1 / q Q-PAM
error amplitude is
(quantized signal
amplitude) limited to  k  1/ q
2/q

PAM (analog signal

amplitude)
time

TS

  q  1 / q

18 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Quantization noise: uniform quantization

 Model the quantized signal by assuming ideal PAM sampling

using the quantization error k:
y(t )    x(kTS )   k  (t  kTS )
k

 Quantization error is the difference of the PAM and the Q-PAM

signals
  x (kT )  x(kT )
k q S S

 The final output is obtained by using the ideal LPF:

yD (t )  x(t )    k sinc( f S t  k )
k

 Assuming equal probable signal at all amplitude levels yields for

quantization noise average
Q-noise PDF
power
q 1/ q 2 1 q/2
 k    d  2
2

2 1/ q 3q
1/ q 1/ q
2q
note: 1
19 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
q2
 PCM versus analog modulation
 PCM suppresses wideband noise  S / N   3D S 
D D FM
2
x

 A PCM system well above threshold  SD  3q 2

 N   1  4q 2 P S X
S D / N D  3q 2 S x  3M 2 b S x , q  M n  D  PCM e

transmission BW: BT  bW channel noise

 Observations:
– operate PCM near threshold!
 Note that for radio 75 dB dynamic
range (12 bits without compression)
transmission bandwidth BT is
W  15kHz, n  12bit
BT  nf s / 2  12  35/ 2  210kHz
(This is about the same as in FM)
 Why to use PCM then??
S x  1/ 2
– regenerative digital repeaters
– digital multiplexing
– Helsinki
etc!University of Technology,Communications Laboratory, Timo O. Korhonen
  S / W x
20
21 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 Some line codes

Unipolar [0,A] RZ and NRZ

Polar [-A/2,A/2] RZ and NRZ

Bipolar [-A/2,0,A/2] AMI

Split-Phase Manchester

Code rate Split-Polar quaternary NRZ

reduced
by n M  2n
rbe  r / n  r / log 2 M

22 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen

 Repeaters
 At the transmission path regenerative repeaters are often used
 At the receiver signal is transformed back to analog form by
lowpass filtering removing harmonics produced by sampling
 Repeaters are categorized as:

– analog repeater: gain equal to the line attenuation

between repeaters
– digital repeater: regenerates bits Analog and digital
by decoding and encoding repeater chains
Error rates for polar, baseband, compared
(polar code)
m-stage repeater chains:
( S / N )1
SD / N D  analog repeaters* Pe  105
m
 Pe  Q  ( S / N )1 / m 
* Formula follows from the
Pe  mQ  ( S / N )1  digital repeaters reverse inspection than
coherent averaging
23 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 PCM encoding and decoding circuits (n=3)
PAM
(from sampler)

always positive sign-bit

irrespective of
x(kTs) polarity
(to parallel-to
serial converter)

quantization
levels

ADC

(from serial-to
parallel converter)
direct-conversion encoder

(to LP-filter)

Q-PAM
=DAC-formula
DAC (inverting adder!)
weighted-resistor decoder
24 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 PCM Systems and Digital Time Division
Multiplexing (TDM)
 In digital multiplexing several messages are transmitted via
same physical channel. For multiplexing 64 kbit/s channels in
digital exchanges following three methods are available:
– PDH (plesiochronous digital hierarchy) (the dominant
method today, E1 & T1) (‘50-’60, G.702)
– SONET (synchronous optical network) (‘85)
– SDH (synchronous digital hierarchy) (CCITT ‘88)

European PCM frame 125  s

32 time slots x 8 bits x 8000 Hz = 32 x 64 kbit/s = 2048 kbit/s

frame synchronization slot

signaling or traffic

traffic PDH E-1 frame

25 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 PCM Hierarchy in PDH
European hierarchy USA hierarchy

139.26 Mbit/s 139.26 Mbit/s

x4 x3
34.368 Mbit/s 44.736 Mbit/s
x4 x7
8.448 Mbit/s 6.312 Mbit/s
x4 x4
2.048 Mbit/s 1.544 Mbit/s
... ...
x32 x24
64 kbit/s 64 kbit/s

If one wishes to disassemble a tributary from the main

flow the main flow must be demultiplexed step by step to
the desired main flow level in PDH.
26 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 E1 and T1 First Order Frames Compared*
T1 is byte-interleaved:
blocks of eight bits from
the same channel
are inserted to the
multiplexed flow

T1 1/125 s  8 kHz
=8000 frames/sec
USA & Number of channels
Japan synchronization

E1 signaling info
Europe

NOTE: In T1 one bit in each time slot in every sixth frame

is replaced by signaling information yielding net flow of
56 kb/s per channel

*John G. van Bosse: Signaling in Telecommunication Networks

27 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen
 T1 and E1 Summarized
 In PSTN two PCM systems dominate:
– T1, developed by Bell Laboratories, used in USA & Japan
– E1, developed by CEPT* used in most of the other countries
 In both data streams divided in frames of 8000 frames/sec
 In T1
– 24 time-slots and a framing (F) bit serves 24 channels
– Frame length: 1+ 8x24=193 bits
– Rate 193x8000 bits/second=1544 kb/s
 In E1
– frame has 32 time-slots, TS 0 holds a synchronization
pattern and TS 16 holds signaling information
– An E1 frame has 32x8=256 bits and its rate us
8000x256=2048 kb/s

*European Conference of Postal and

Telecommunications Administration
28 Helsinki University of Technology,Communications Laboratory, Timo O. Korhonen