CP1_7A_note

8/3/2019 CP1_7A_note

1/21

HKIVE (ST)

Dept. of Electronic and Information Engg. Page 1 Comm. Principles 1 / Chapter 7

TELEPHONY

Public telecommunication networks are originally

designed for telephony and are known as Public

Switched Telephone Network (PSTN). They aredigitized to provide a variety of teleservices and are

known as Integrated Services Digital Network (ISDN).

1 Network Structure

Since voice communications take minutes to finish one

call, star configuration is the best choice in term of cost.

This is applied to connection between telephone

terminals and local exchanges. The connections

between exchanges have a 'tree' modified hierarchical

structure. The exchange centers are classified into

different levels which depend on their sizes and

locations (Fig. 1.1, 1.2)

Primary Centers

Secondary Centers

Local Exchanges

Tertiary Centers

Fig. 1.1 Hierarchical Structure Conform to ITU-T

Recommendation

8/3/2019 CP1_7A_note

2/21

HKIVE (ST)


Fig. 1.2 Example of Hierarchical Switching Networks

1.2 Standards

Standardization of international telecommunication is

carried by two sectors of International

Telecommunications Union (ITU), namely:

ITU Telecommunications Sector (ITU-T). Its duties

include the study of technical questions, operating

methods and tariffs for telephony, telegraphy and data

communications.

ITU Radiocommunication Sector (ITU-R). Its duties

include the study of technical questions, operating

methods and tariffs for radio communications.

A telecommunications network consists of the

following interacting subsystems:

(i) Transmission systems

(ii) Switching systems

(iii) Signalling systems

8/3/2019 CP1_7A_note

3/21

HKIVE (ST)


2 Transmission Systems

Basic telephone service involves the transmission of

speech to a distance place. In analogue system, atelephone line has a 300 to 3.4 kHz bandwidth mainly

for speech communication. PCM method is used to

convert analog speech to digital voice. They are then

time-division-multiplex TDM to form high capacity

trunk for transmission and switching.

2.1 Pulse Code Modulation (PCM)

It is an analog-to-digital conversion process that

converts analog signal into a format compatible with

digital transmission and switching.

Anti-

filter

Analogsignalaliasing Compressor

Encoder

PCMsignalQuantizer

and A/DSampling

Line

Fig. 2.1 PCM Generation

8/3/2019 CP1_7A_note

4/21

HKIVE (ST)


Sampling

It is a conversion of analog waveform into pulse

amplitude modulation (PAM) form where the amplitude

of the pulse denotes the analog amplitude at thesampling time.

t

t

t

Analog waveform

Sampling Pulse

Resulting PAMsignal

fs

1

Fig. 2.2 Sampling of Analog Signal

Fig. 2.3 shows the spectrum of the PAM. Fig. 2.4

shows that if the sampling rate is least than the double

of the signal bandwidth, distortion will result and this

distortion is called aliasing. Therefore, an anti-aliasing

filter is added before sampling to safe-guard the bandwidth. This minimum sampling rate is known as

the Nyquist sampling rate:

fs = 2W (2.1)

8/3/2019 CP1_7A_note

5/21

HKIVE (ST)


where W= bandwidth of the signal and

fs = sampling rate

Fig. 2.3 Spectrum of (a) the Modulating Signal and

(b) the PAM Wave

Fig. 2.4 (a)fs > 2W(b)fs < 2W

Quantization

It is the process of rounding off the amplitudes of thesamples to certain predetermined levels available to the

A/D converter.

8/3/2019 CP1_7A_note

6/21

HKIVE (ST)


In Fig. 2.5(a), the straight line shows the linear input

and output relationship, and the staircase function

shows the quantized relationship. Fig. 2.5(b) shows the

quantization error as a function of input voltage. Thequantization error appears as noise after recovery and is

referred to as quantization noise.

The quantization error can lie between V/2, and

assuming it has a uniform probability density

distribution, the rms quantization error (noise) is

Enq =12

Vsee appendix (2.2)

where V = step size (resolution)

More generally, the ratio between the peak and rmsvalues of the signal voltage will be some value k =

Erms/Emax. If distortion is to be avoided, the peak signal

level must not be allowed to exceed the peak input of

the quantizer.

Emax = 2

LV

where L = total number of steps

The signal-to-quantization noise ratio in this case is

(S/N)q = Erms2/Enq2 = 3k2L2 (2.3)

8/3/2019 CP1_7A_note

7/21

HKIVE (ST)


Fig. 2.5(a) Linear Quantization; (b) Quantization Error

8/3/2019 CP1_7A_note

8/21

HKIVE (ST)


Sign-Magnitude Code

The MSB of the 8-bit codeword is the sign bit and the

remaining bits are the magnitude bit. The example

below show a 3-bit PCM code:

Sign Magnitude

1 11

1 10

1 01

1 00

0 00

0 01

0 10

0 11

0.5V

1.5V

2.5V

3.5V = Emax

-0.5V

-1.5V

-2.5V

-3.5V

Transfer Function Recovered

Voltage

3 V

2 V

1 V

0 V

0 V

-1 V

-2 V

-3 V

V

Fig. 2.6 Staircase Transfer Curve

From Fig. 2.6, it clearly shows that the total number ofstep L:

L = 2n - 1 where n = no of bit per codeword (2.4)

Companding

With speech the peaks of the signal only infrequently

extend over the full range of the input. In effect, the

8/3/2019 CP1_7A_note

9/21

HKIVE (ST)


signal does not have a uniform probability density

function, and the (S/N)q is lower than that given by Eq.

2.3. To compensate for this, compression is needed. It

has variable gain characteristics with a lower gain at

higher input.

If we keep the bits per sample fixed, then compression

will increase the S/Nq for low amplitude signals and

will decrease S/Nq for higher amplitude signals as

compared with the linear PCM. Ideal companded PCM

encodes signal levels with quantization error

proportional to signal level to keep the S/Nq constant atall amplitude levels.

In North America, -law compression characteristic is

used. In Europe and many other parts of the world, A-

law characteristic is used. They all use 8-bit

quantization.

Fig. 2.7 Compressor Characteristics of and A-law

8/3/2019 CP1_7A_note

10/21

HKIVE (ST)


Fig. 2.8 Nonlinear Quantization

The reverse of compression is expansion. The

combined process is known as companding.

Some Important Data

With 3.4 kHz being the maximum voice frequency in

telephony, the standard sampling rate selected is

8,000 samples per second. With 8-bit quantizer, the

overall PCM data rate is 64 kbps.

8/3/2019 CP1_7A_note

11/21

HKIVE (ST)


Line Encoding

It is used to convert standard logic levels to suitable

format for telephone line transmission. The followings

are some examples of line encoding method:

Line Encoding

Unipolar NRZ

Bipolar NRZ

Unipolar RZ

Bipolar RZ

AMI

0 0 1 1 0 1 0 1

0 V

+2 V

+1 V

-1 V

0 V

+2 V

0 V

-1 V

+1 V

0 V

-1 V

+1 V

NRZ = Non-Return-to-Zero

RZ = Return-to-Zero

AMI = Alternate Mark Inversion

Fig. 2.9 Line Encoding Methods

8/3/2019 CP1_7A_note

12/21

HKIVE (ST)


Major factors in selecting a line encoding format:

Transmit Power

Unipolar uses more power than the bipolar methods.

DC Component

Unipolar methods have DC components whilst bipolar

methods have not. The presence of DC in the signal

does not favour the use of transformer for isolation.

Clock Recovery

Bipolar RZ has the highest voltage transitionswhich is the best for clock recovery.

Unipolar RZ and AMI lack of transitions for longstrings of zero.

NRZ lacks of transitions for long strings of zeroand one.

Bandwidth

The bandwidth of RZ doubles to that of NRZ and AMI.

Error Detection

AMI possesses error detection ability while all others

do not.

The conclusion is that AMI is the best line encodingmethod among all others in our example.

8/3/2019 CP1_7A_note

13/21

HKIVE (ST)


2.2 Time-Division Multiplexing

In the European digital hierarchy, it uses the ITU-T 30

PCM voice channels to form the primary multiplex

carrier and has a data rate of 2.048Mbps.

30-channel Frame Format

Fig. 2.10 shows the ITU-T 2.048 Mbps

recommendation. Sixteen frames form a multi-frame

and in each frame, there are 32 eight-bit time slots, 30

for PCM voices, one for synchronization and one for

channel signalling.

1 2 3 4 13 14 15 16 17 18 19 28 29 30 310 0 1

1 3 5 72 4 6 8 8-bit PCM

One frame, 256 bits, 125usecs

ch 1 2 .... 15 16.... 17 18 .... 30....

Synchronizationchannel

Next frame

Signalling

channel

Fig. 2.10 30 Channels PCM Multiplexing

One frame contains one sample per user, therefore

Sampling rate = Frame rate = 8k per second

Frame duration = 1/8k = 125 s

Total number of bit per frame = 32 8 = 256 bits

Multiplex data rate = 256 /125 s = 2.048 Mbps

8/3/2019 CP1_7A_note

14/21

HKIVE (ST)


Plesiochronous Digital Hierarchy (PDH)

Higher-order digital multiplex systems are first based

on PDH since all the primary inputs operate from

independent clock sources. It operates at 2.048M,8.448M, ... , multiplex hierarchy.

Synchronous Digital Hierarchy (SDH)

It is for fully digitized networks operating

synchronously using high-capacity optical-fiber

transmission systems and TDM switching. It operates

at basic rate of 155.52Mbps and multiplexes by factor

of 4. Any of the existing plesiochronous rates up to

140Mbps can be multiplexed into the SDH.

24-channel Frame Format

The other common multiplex format is the 24-channel

multiplexing used in USA.

2 3 4 5 10 11 12 13 14 15 16 21 22 23 241 1 2

1 3 5 72 4 6 8 8-bit PCM

One frame, 193 bits, 125usecs

Synchronization Bit

Next frame

Fig. 2.11 24 Channels PCM Multiplexing

Total number of bit per frame = 24 8 + 1 = 193 bits

Multiplex data rate = 193 /125 s = 1.544 Mbps

8/3/2019 CP1_7A_note

15/21

HKIVE (ST)


3 Switching

3.1 Single-stage Space Switch

It is a simple full-matrix switch. However, a largeswitch system cannot be formed simply using full-

matrix switches because the number of crosspoints N2 is

huge when N is large.

1

2

N

1 2 N

N xN::

....

Fig. 3.1 (a) Matrix switch (b) Symbol

Concentrator and Expander

N xN N xN N xN. . . . .

1 2 M

N M x N

(a)

NM x N

Concentrator Expander (b)

Fig. 3.2 (a) Concentrator and Expander using SquareMatrix Switches; (b) Symbolic Representations

8/3/2019 CP1_7A_note

16/21

HKIVE (ST)


3.2 A General Trunking

In order to utilize the switching network more

effectively, the low-usage subscriber lines are grouped

into high-usage trunk-groups before switching.

Switching

ConcentratorSubscriber

lines

Expander

IncomingJunctions

OutgoingJunctions

Central control

Fig. 3.3 General Trunking for a Switching System

3.3 Two-stage Space Switch Network

To reduce the total number of crosspoints, multi-stage

switch network is constructed using smaller matric

switches. The method is illustrated in Fig. 3.4 with a

10 10 matrix switch as the basic building block.

The basic design criterion is that the number of link

(intermediate connections) should be equal to the

number of trunk (incoming or outgoing connections).

This ensures the utilization of links will not be under or

over to that of the trunks. This leads to:

8/3/2019 CP1_7A_note

17/21

HKIVE (ST)


The matrix switch should be square, i.e., N N The optimum N = Trunkof. No (3.1)

10x 10

10x 10

10x 10

10x 10

(10) (10)100 incoming 100 outgoing

trunks trunks

link

Fig. 3.4 Two-stage Switch Network

The blocking probability is high since there is only one

link between any two input and output building blocks.

3.4 Three-stage Space Switch Network

Three-stage network resolves or reduces network

blockage by the introduction of additional stage in the

middle. In Fig. 3.5, there are ten intermediate links

instead of one in the 2-stage network.

10x10

(10)100 incoming

10x10

trunks

10x 10

(10) 100 outgoing

10 x 10

trunks

10x10

(10)

10x10

Fig. 3.5 Example of 3-stage Switch Network

8/3/2019 CP1_7A_note

18/21

HKIVE (ST)


3.5 Time Switching

If TDM transmission is used with space switching (Fig.

3.6), it is necessary to provide de-multiplexing and

multiplexing equipment before and after switching. Iftime switching is employed (Fig. 3.7), the multiplex

links can be switched directly.

Space-

division

switch

Incoming

PCM

trunks

PCM

muldexes

Outgoing

PCM

trunks

PCM

muldexes

Fig. 3.6 Space Switching with PCM

Time-

division

switch

Outgoing

PCM

trunks

Incoming

PCM

trunks

Fig. 3.7 Time Switching with PCM

The principle of a time switch is shown in Fig. 3.8.

Since any incoming channel can be connected to anyoutgoing channel, it is equivalent to an N N matrix

switch.

8/3/2019 CP1_7A_note

19/21

HKIVE (ST)


0

1

2

3

X

Cyclic

write

From input

trunk

Decoder

Speech

store

Connection

store

S p e e c h

address of x

0

1

2

3

Y

Random

read

indirectly

To output

trunk

Fig. 3.8 Operation of a Time Switch

Time switching and space switching are used togetherin tandem switch and are known as Time-Space-Time

(TST) switching (Fig. 3.9).

m outgoingtrunksm incomingtrunks

Time

switch

nn

Time

switch

nn

Space

switch

mm

Time

switch

nn

Time

switch

nn

Fig. 3.9 TST Switch Networkm = no. of PCM trunks,

n = no. of timeslots.

8/3/2019 CP1_7A_note

20/21

HKIVE (ST)


Time switch doesn't have the size problem when

compared with space switch. It can be expanded

rapidly by increasing the memory size and the operating

speed. However, time switch has the problem of

delay. Each time switch will introduce at most 125 sdelay time.

4 Signalling

To maintain smooth operations of a telephone network,

there are a lot of signalling information needed to be

sent through the telephone network. When the signals

are sent along with the same voice circuits, this is

known as channel-associated signalling. However, it

is more efficient for the central processor of one

exchange to send information to the other exchange

through separate high-speed data link, and this is known

as common-channel signalling. An example is thetimeslot-16 of European primary muliplex. It provides

a 64 kbps common-channel for signalling between

processors.

8/3/2019 CP1_7A_note

21/21

HKIVE (ST)

Appendix

From Fig. 2.13(b), we find the relationship to describe

the error voltage:

y =2

V- x for 0 < x

CP1_7A_note

Documents