5i'
PART II
SUPPLEMENTS TO SECTION 1
OF THE SERIES G RECOMMENDATIONS
Blanc
MONTAGE: PAGE PAIRE = PAGE BLANCHE
Supplement No. 1
CALCULATION OF THE STABILITY OF INTERNATIONAL CONNECTIONS 
ESTABLISHED IN
ACCORDANCE WITH THE TRANSMISSION AND SWITCHING PLAN
(Referred to in Recommendation G.131; this Supplement is to be found
on page 555 of Volume III.2 of the Green Book , Geneva, 1973)
Supplement No. 2
TALKER ECHO ON INTERNATIONAL CONNECTIONS
(Geneva, 1964; amended at Mar del Plata, 1968 and Geneva, 1976 and
1980;
MalagaTorremolinos, 1984; and Melbourne, 1988;
referred to in Recommendation G.131, S 2)
1 The curves of Figure 2/G.131 may be used to determine
whether a given international connection requires an echo control
device (echo suppressor or echo canceller). Alternatively they may
be used to find what value of nominal overall loss shall be adopted
for the 4wire chain of a complete connection so that an echo con
trol device is not needed. Before the curves can be used it must be
decided what proportion of calls are to be allowed to exhibit an
objectionable echo and Recommendation G.131 gives guidance on this
matter.
The coordinates of the graph represent two of the parameters
of a telephone connection that govern echo, i.e. the overall loud
ness rating (OLR) of the echo path and the mean oneway propagation
time. By making certain assumptions (discussed below) these two
parameters become the principal ones.
Each curve divides the coordinate plane into two portions and
the position, relative to the curve, of the point describing the
connection indicates whether an echo control device is needed,
bearing in mind the percentage of calls permitted to exhibit an
objectionable echo.
2 Factors governing echo
The principal factors which must be considered in order to
describe whether an echo control device is needed on a particular
connection are:
a) the number of echo paths;
b) the time taken by the echo currents to traverse
these paths;
c) the OLRs of the echo path including the sub
scriber lines;
d) the tolerance to echo exhibited by subscribers.
These factors are discussed in turn in the following.
When circuits are switched together 4wire there is only one
echo path, assuming negligible gotoreturn crosstalk. This is also
substantially true if the circuits are switched together 2wire and
good echo return losses are achieved at these connection points
(e.g. a mean value of 27 dB and a standard deviation of 3 dB). The
principal echo currents are those due to the relatively poor echo
balance return losses at the ends of the two extreme 4wire cir
cuits, where the connection is reduced to 2wire.
The time taken to traverse the echo path is virtually depen
dent solely on the length of the 4wire connection, because the
main circuits of modern national and international networks are
highvelocity circuits.
The OLR of the talker echo path for a symmetrical connection
for planning purposes is approximately given by the sum of:
 twice the junction loudness rating (JLR) of the
connection between the 2wire point in the talker's local terminal
exchange and the 2wire side of the 4wire/2wire terminating set
at the listener's end ;
 the echo balance return loss at the listener's
end;
 the sum of the sending LR and receiving LR of the
talker's telephone and subscriber line;
In general, values of sending LR and receiving LR correspond
ing to lowloss subscriber lines should be used.
The echo experienced by subscribers on lines with more loss
will be further attenuated. This is, therefore, a conservative
assumption.
The data on tolerance to echo exhibited by subscribers given
in Table 1 are furnished by the American Telephone and
Telegraph Co. and are based on a series of studies completed in
1971. These tests provided information on the overall loudness rat
ing (EARS) of the echo path for echo, just detectable, as a func
tion of echopath delay. In addition, ratings of quality on a
fivepoint scale (excellent, good, fair, poor, unsatisfactory) were
also obtained. The values in terms of EARS loudness ratings (then
used by AT&T) were subsequently translated to values of CCITT loud
ness ratings by adding 1 dB. Table 1 indicates the mean echo path
loss for the threshold of detectability and for ratings of unsatis
factory. These mean values are the loudness rating of the echo path
for 50% detectability and 50% unsatisfactory. The standard devia
tion is also given.
H.T. [T1.2]
TABLE 1
Results of echo tolerance tests
_________________________
According to Recommendation G.111, S A.3.3 the Junction
Loudness Rating of 4wire circuits should be taken as
the 800 or 1000 Hz loss.
_____________________________________________________________________________
{
{
Threshold Unsatisfactory Mean (dB) Standard deviation (dB)
_____________________________________________________________________________
10 26 = 4 9 = 6
20 35 = 4 16 = 6
30 40 = 4 20 = 6
40 45 = 4 23 = 6
50 50 = 4 25 = 6
75  =  29 = 6
100  =  32 = 6
150  =  35 = 6
200  =  37 = 6
300  =  39 = 6
_____________________________________________________________________________













































































































TABLE 1 [T1.2] p.
3 Construction of Figure 2/G.131
The mean margin against poor or unsatisfactory echo perfor
mance is given by:
M = 2T + B  E + SLR + RLR
where
T is the mean junction loudness rating of the connection
between the 2wire point in the talker's local terminal exchange
and the 2wire side of the 4wire/2wire terminating set at the
listener's end. The loudness rating is assumed to be the same in
both directions of transmission;
B is the mean echo balance return loss at the
listener end;
E is the mean value of loudness rating of the echo
path required for an opinion rating of unsatisfactory ;
SLR is the sending loudness rating at the 2wire
_________________________
This corresponds to the value of overall loudness rat
ing of the echo path at which 50% of the opinion rat
ings are unsatisfactory.
point in the originating local exchange for short subscriber lines;
RLR is the receiving loudness rating at the 2wire
point in the originating local exchange for short subscriber lines.
4 Fully analogue connections
The echo balance return loss is assumed to have a mean value
of not less than 11 dB, with a standard deviation of 3 dB expressed
as a weighted meanpower ratio (see Recommendation G.122). The mean
value of the transmission loss is assumed to be uniform over this
band and the standard deviation of transmission loss for each
4wire circuit is assumed to be 1 dB for each direction of
transmission. The correlation between the variations of loss of the
two directions of transmission is assumed to be unity.
The standard deviation of the margin is given by:
m 2 = n  t 2
1 + 2rt1t2
+
t 2
2) + b 2 + e 2
where
m is the standard deviation of the margin;
t1, t2 are the standard deviation of the transmission loss
in the two directions of transmission of one 4wire circuit,
national or international;
b is the standard deviation of echo balance return
loss;
e is the standard deviation of the distribution of talker
echo path loudness ratings required for opinion ratings of unsatis
factory;
r is the correlation factor between t1and t2;
n is the the number of 4wire circuits in the 4wire
chain.
Inserting t1 = t2 = 1 dB; r = 1; b = 3 dB; e = 6 dB gives m
2 = (4n + 45).
In Recommendation G.131, S 2.3, Rules A and E refer to 1% and
10% probabilities of encountering unsatisfactory echo and for these
cases nine 4wire circuits are assumed (3 national and 3 interna
tional + 3 national). For both the 1% and 10% curves therefore m
= 9.0 dB.
For 10% probability, the margin may fall to 1.28 times the
standard deviation. The corresponding factor for the 1% curve is
2.33. Hence the corresponding values of M are:
M = 1.28 x 9.0 dB = 11.5 dB for 10% probability
M = 2.33 x 9.0 dB = 21  dB for 1% probability.
Putting these values into M = 2T + B  E + SLR + RLR gives
the following values for the mean talker echo attenuation, 2T + B
+ SLR + RLR:
2T + B + SLR + RLR = 11.5 dB + E for 10% probability
2T + B + SLR + RLR = 21  dB + E for 1% probability.
The values in Table 2 have been calculated (to the nearest
whole decibel) using these equations. The figures in the length of
connection column have been calculated assuming a velocity of pro
pagation of 160 km/ms.
H.T. [T2.2]
TABLE 2
_______________________________________________________________
{
{
10% unsatisfactory 1% unsatisfactory
_______________________________________________________________
10 1  00 21 30
20 3  00 28 37
30 4  00 32 41
40 6  00 35 47
50 8  00 37 46
75 12  00 41 50
100 16  00 44 53
150 24  00 47 56
200 32  00 49 58
300 48  00 51 60
_______________________________________________________________




























































































































TABLE 2 [T2.2] p.
The solid last line for n = g in Figure 2/G.131 has been con
structed from these values and similar values calculated for other
values of n (analogue circuits).
5 Fully digital connections with analogue 2wire subscribers
lines (conform to Figure 2/G.111)
The standard deviation of the margin is given by:
m 2 = n (2t 2
) + b 2 +
e 2
where
m is the standard deviation of the margin;
n is the number of coder/decoder pairs;
t is the standard deviation of the transmission loss in
the two directions of transmission;
b is the standard deviation of the echo balance
return loss;
e is the standard deviation of the distribution of
talker echo path loudness ratings required for opinion ratings of
unsatisfactory.
The term (2t 2 ) represents the loss variance of a
coder/decoder pair, where t = 0.2 dB. For a fully digital connec
tion between 2wire analogue subscribers lines there are
2 coder/decoder pairs (i.e. one at the talker's local exchange and
one at the listener's local exchange).
Inserting n = 2, t = 0.2 dB, e = 6 dB and assuming b = 3 dB
gives m 2 = 45.2 and m = 6.7.
Hence the values of m are:
m = 1.28 x 6.7 dB = 8.6 dB for 10% probability
m = 2.33 x 6.7 dB = 15.6 dB for 1% probability.
Putting these values into M = 2T + B  E + SLR + RLR gives the
following values for the mean talker echo path attenuation, 2T + B
+ SLR + RLR:
2T + B + SLR + RLR = 8.6 dB + E  for 10% probability
2T + B + SLR + RLR = 15.6 dB + E  for 1% probability.
The values in Table 3 have been calculated using these equa
tions.
H.T. [T3.2]
TABLE 3
_____________________________________________________________
{
{
10% unsatisfactory (dB) 1% unsatisfactory (dB)
_____________________________________________________________
10 17.6 24.6
20 24.6 31.6
30 28.6 35.6
40 31.6 38.6
50 33.6 40.6
75 37.6 44.6
100 40.6 47.6
150 43.6 50.6
200 45.6 52.6
300 47.6 54.6
_____________________________________________________________



























































































Table 3 [T3.2], p.
The dashed line in Figure 2/G.131 has been constructed from
these values (fully digital connections).
6 Fully digital connections with digital subscribers lines
(conform to Recommendation G.801)
In this case there are no 2wire points in the connection.
However, there is an acoustical feedback path between the earpiece
and mouthpiece of the telephone set. Therefore the echo balance
return loss used above is now represented by the loss of this
acoustical path. Representative values of this acoustical loss are
under wider study. The appendix to this supplement gives some
information on this question.
It may be assumed that the standard deviation of the transmis
sion loss of the coder/decoder pair equals the value given above
for digital connections with 2wire subscriber lines. The value of
the equivalent of T should be taken as zero. The quantities SLR and
RLR now refer to virtual analogue 4wire points of 0 dBr level.
If it can be assumed that the standard deviation of the
acoustical echo path loss equals 3 dB and a normal distribution
applies, then the values of Table 3 also apply to the digital sub
scriber line case and the dashed curve of Figure 2/G.131 may be
used.
7 Mixed analogue/digital connections
This case is a combination of the cases given above and the
appropriate variables and their values should be taken from the
above information and an appropriate table can be constructed.
In general, if there are only two coder/decoder pairs in the
connection, the variability of the transmission loss of the codecs
may be ignored compared with the variability of the analogue cir
cuits and the other variabilities. For such connections the solid
curve given in Figure 2/G.131 for the number of analogue circuits
in the connection may be used with good accuracy.
APPENDIX I
(to Supplement No. 2)
Echo loss in 4wire telephone sets
(Contribution by Norway)
Abstract
In a 4wire telephone set, echo may arise both by electrical
crosstalk in the cord and by acoustical coupling between earpiece
and mouthpiece in the handset. The echo loss for these paths has
been determined for two analogue 2wire telephone sets. This data
is used to derive the echo loss of a hypothetical 4wire set having
SLR + RLR = 3 dB, and acoustical and electrical properties the same
as the 2wire telephone sets.
I.1 Introduction
It has been pointed out in several contributions that the
choice of LRs for digital telephone sets has to be made considering
aspects of loudness and echo in a complete 4wire connection. To
enable a study of the risk of objectionable echo, Study Group XVI
has asked Study Group XII to present information on the subjective
effect of talker echo as a function of delay, overall LR and echo
path loss.
In a digital 4wire connection, including 4wire subscriber
lines and digital telephone sets, the main echo paths are found in
the telephone set itself:
 the acoustical coupling between earpiece and
mouthpiece of the handset, and
 the electrical coupling in the flexible cord to
the handset.
In order to assess the echo performance of a 4wire connec
tion, the echo loss of the digital telephone set must be estimated.
As an example of what can be expected, measurements of these
echo paths have been made on two different analogue telephone sets
results have been used to derive the echo loss between the receive
and send terminals of a hypothetical 4wire telephone set having
SLR = 6 dB and RLR = 3 dB, and having the same electrical and
acoustical properties as the measured sets.
I.2 Measurements
Figure I1 shows a setup for measurement of the loss between
the receive and send direction in an ordinary 2wire telephone. Two
telephone sets are used to separate the two directions of transmis
sion.
The acoustical path is measured by replacing the cord of the
handset by shielded wires and the electrical path is measured by
replacing the microphone by an appropriate resistor. When measuring
the acoustic coupling , the handset was placed both in "free field"
and held in a normal listening position.
Two different Norwegian standard telephone sets were included
in the measurements. Both sets are equipped with linear micro
phones. EB model 67 has a "conventional" handset whereas Testafon
is a "modern" set.
I.3 Echo loss results
In order to enable comparison of the data obtained for the two
telephone sets, the measurements have been referred to a telephone
set having SLR + RLR = 3 dB. The echo loss, as defined in Recommen
dation G.122, S 2.2, for this hypothetical telephone is shown in
Table I1.
The acoustical conditions refer to:
1) the handset held in a normal listening position,
tightly against the ear ("real ear"), and
2) the handset held in "free field".
Figura I1, p.
H.T. [T4.2]
TABLE I1
Echo loss in dB of hypothetical telephone set
having SLR + RLR = 3 dB
______________________________________________________________________
EB model 67 Tastafon
Acoustical condition Free field Real ear Free field Real ear
______________________________________________________________________
Acoustical path 28.2 31.7 41.5 44.4
Electrical path 32.2 32.2 37.0 37.0
______________________________________________________________________


















Cuadro I1 [T4.2], p.
I.4 Discussion
It should be noted that high echo loss has not been design
objective for either of the measured telephone sets. The results
may therefore be considered as representative of what may be
obtained when no special precautions are taken.
The echo loss of the acoustical path is apparently highly
dependent on the physical design of the telephone handset and of
the acoustical properties of the transducers. A difference of 13 dB
is obtained in Table I1 between the two sets in the test. The
effect of acoustical termination of the earphone, i.e. "free field"
or "real ear", is fairly small, approximately 3 dB for both sets.
Table I1 shows that the electrical crosstalk in the flexible
cord is an important echo source in both sets. For a given SLR and
RLR , the level of crosstalk will depend on the partitioning of the
gain between the handset (i.e. the microphone) and the telephone
apparatus. Increasing the gain in the handset (by increasing the
microphone sensitivity or by placing the microphone amplifier in
the handset) will increase the signal level in the cord and improve
the signaltocrosstalk level. The crosstalk may also be reduced by
using shielded wires eliminated by proper design, and the acousti
cal component may be considered as the lower limit for the echo
loss.
Supplement No. 3
EVALUATION OF ECHO CONTROL DEVICES
(Referred to in Recommendation G.161; this Supplement is to be found
on page 559 of Volume III.2 of the Green Book , Geneva, 1973)
Supplement No. 10
APPLICATION OF RECOMMENDATION B.4 
CONCERNING THE USE OF DECIBEL
(This Supplement is to be found on page 598 of Volume III.2 of
the Green Book , Geneva, 1973)
Supplement No. 20
POSSIBLE COMBINATIONS OF BASIC TRANSMISSION IMPAIRMENTS
IN HYPOTHETICAL REFERENCE CONNECTIONS
(Referred to in Recommendation G.103; this Supplement is to be found
on page 319 of Fascicle III.1 of the Red Book , Geneva, 1985)
Supplement No. 21
THE USE OF QUANTIZING DISTORTION UNITS IN THE PLANNING
OF INTERNATIONAL CONNECTIONS
(Contribution of BellNorthern Research)
(Referred to in Recommendation G.113; this Supplement is to be found
on page 326 of Fascicle III.1 of the Red Book , Geneva, 1985)
Supplement No. 24
CONSIDERATION CONCERNING QUANTIZING DISTORSION UNITS
OF SOME DIGITAL DEVICES THAT PROCESS ENCODED SIGNALS
(Referred to in Recommendation G.113; this Supplement is to be found
on page 333 of Fascicle III.1 of the Red Book , Geneva, 1985)
Supplement No. 25
GUIDELINES FOR PLACEMENT OF MICROPHONES AND LOUDSPEAKERS
IN TELEPHONE CONFERENCE ROOM
(Referred to in Recommendation G.172; this Supplement is to be found
on page 335 of Fascicle III.1 of the Red Book , Geneva, 1985)
Supplement No. 29
OBJECTIVE FOR THE MIXED ANALOGUE/DIGITAL CHAIN OF  fR 4WIRE
CIRCUITS
Draft Recommendation G.136
(This Supplement is proposed for further study during the
present study period with the aim to convert the supplement into a
Recommendation.)
1 General
In the period of transition from a fully analogue to a fully
digital network, there will be, on international and national net
works, mixed type chain of 4wire telephone circuits (see
Recommendation G.101, S 4.2), some sections of which can be made
with analogue or digital transmission systems.
Considering the fact that the transition period may last for a
fairly prolonged time, and also considering the need for guarantee
ing a certain quality of transmission on mixed chain of circuits,
the CCITT recommends observance of some principles for the composi
tion of mixed chain of circuits as set forth below and some objec
tives for their parameters.
The main principle in the standardization of mixed circuits
lies in the retaining of the standards adopted for the FDM cir
cuits. This would have resulted in retaining the transmission qual
ity over the 4wire chain formed by the international circuits and
national extension circuits.
For some parameters this can be achieved, but as far as some
other parameters are concerned due to analogue/digital conversions
and errors in digital sections there are some considerable differ
ences in standards and measuring methods.
Objectives for some mixed circuit parameters are contained in
a number of G, Q, and Mseries Recommendations. However, these
objectives do not take due account of the addition laws for distor
tions based on the multitude of mixed circuit structures and
specific features of the measuring methods involved.
Considering the importance of retaining the transmission qual
ity during the transition period and attaching great importance to
the standardization of mixed analogue/digital circuits the multitu
dinous types of which emerge while using various kinds of
analoguetodigital conversions, CCITT thinks it worth while to
have a specific Recommendation on objectives for mixed
analogue/digital circuits and 4wire chains including both analogue
and digital circuits.
The present Recommendation related to mixed 4wire chain of
circuits and the analogue/digital mixed connections dealt with in
this Recommendation are those with analogue telephone sets at both
ends.
It is based on the existing Recommendations for FDM channel
equipment G.232, for PCM channel equipment G.712, for analogue
switching centres Q.45, Q.45  fIbis , for digital switching
centres Q.551 to Q.554, and takes account of other existing Recom
mendations of G and Mseries.
Later on in accordance with the study results of
Question 26/XII the present Recommendation will have to be suppli
mented by objectives for mixed chain of circuits formed with the
help of various methods of analoguetodigital conversion such as
transmultiplexers (Recommendations G.793, G.794), modems
(Recommendations G.941, V.37), transcoders (Recommendation G.761),
group codecs (Recommendation G.795), DCME, as well as connections
with a digital telephone at one end and an analogue telephone at
the other end.
2 Structure of a mixed analogue/digital voice frequence chain
of 4wire circuit
The parameters of a mixed 4wire chain are essentially depen
dent on the number of analogue sections and on the number of
analogue/digital conversions in the chain.
According to Recommendation G.103 the total number of 4wire
circuits in a 4wire chain of the maximum length is 12 in
exceptional cases (Table 2/G.101) so that it may be assumed that
the number of circuits will not exceed 12. The worst cases in terms
of distortions occur when:
 all switching centres are digital, and the cir
cuit sections from and to the centres are set up on analogue
transmission systems. The number of analogue/digital conversions is
then 11, the number of analogue sections is 12;
 all switching centres are analogue, and the cir
cuit sections from and to the centres are set up on digital sys
tems. The number of analogue/digital conversions is 12 in this
case, the number of digital sections is 12.
Such cases are very rare. More representative is considered to
be a case where the number of analogue/digital conversions makes
one half of the maximum number (Recommendation G.103, Annex B),
that is 6, and digital islands are available. The structure of such
a 4wire chain is presented in Figure 1. The number of analogue
sections is 6, the number of digital sections is also 6. Other
structures of mixed 4wire chain come into the picture when connec
tion of the sections is realized without a switching equipment.
These structures are considered in Recommendation M.562 (S 3.2).
The worst case for a circuit of 12 sections without switching cen
tres occurs when digital and analogue sections alternate (see Fig
ure 2), the number of analoguedigital conversions being equal
to 6, the number of digital sections to 6, and the number of analo
gue sections also to 6.
Figure 1, p.
Figure 2, p.
Thus, the examination of various structures of mixed
analogue/digital voicefrequency chain of circuits shows that for a
4wire chain of maximum length having 12 sections, it is advisable
to establish objectives of distortions based on 6 analogue/digital
conversions, 6 analogue and 6 digital sections.
Intermediate variants for combinations of analogue, digital
sections and analoguetodigital conversions will be:
11 analogue sections + 1 a/d conversion
(1 digital section) = 12
6 analogue sections + 6 a/d conversions
(6 digital sections) = 12
It should be borne in mind that the chains may most frequently
consist of less than 12 sections. The contribution of switching
centres to distortion is negligible, if they do not contain
analogue/digital conversions.
3 Objectives for parameters of mixed analogue/digital cir
cuits
3.1 The nominal value of the input/output impedance of the
analogue and digital sections and of a switching equipment should
be 600 ohms.
3.2 Return loss of the input/output impedance referred to the
nominal value of the analogue and digital sections and of a switch
ing equipment should preferably be not less than 20 dB in the
3003400 Hz band.
Note  For a switching centre and channel FDM equipment, the
value of 15 dB is permissible in the 300600 Hz band (see
Recommendation Q.45, S 6.3 and Recommendation G.232, S 7).
3.3 Unbalance loss in respect to earth
The existing Recommendations for switching centres
(Q.45, Q.553) and channel FDM equipment (G.712) standardize the
unbalance loss in respect to the earth in different ways. There are
differences in the measuring methods as well. The Recommendation
for the FDMchannel equipment (G.232), does not specify this param
eter. The question of standardization and methods of measuring this
parameter for mixed circuits channels is under study.
Pending the establishment of unified objectives and measuring
methods, Recommendation K.10 on the unbalance loss of communication
equipment should be referred to in general guidelines in the case
of mixed chain of 4wire circuits.
3.4 Nominal relative level
The nominal relative level on the transmit side of each sec
tion (analogue and digital) is 14 (16) dBr. The nominal relative
level on the receive side of each section (analogue and digital) is
+4 (+7) dBr (see Recommendations G.232, S 11, G.712, S 14, Q.45,
S 3 and Q.553 S 2.2)
The nominal relative level at the virtual analogue switching
point is
 sending: 3.5 dBr
 receiving: 4.0 dBr for analogue
3.5 dBr for digital
(See Recommendation G.101, S 5.2.)
The nominal relative value in a mixed circuit is defined for a
frequency which is not a subharmonic of the sampling frequency. The
recommended tentative value for the frequency is 1020 Hz.
3.5 Variations of transmission loss with time
The standard deviation of the transmission loss should not
exceed 1 dB.
The difference between mean and nominal value of the transmis
sion loss should not exceed 0.5 dB.
Note  The indicated values are defined in Recommendation
G.151, S 3 for a fully analogue circuit under the condition that
the channels are part of a single group equipped with automatic
regulation.
For mixed chains the stability conditions improve on the one
hand because of the existence of digital sections which have a
higher stability than analogue ones; but on the other hand in the
mixed circuits there is no possibility of a transit automatic regu
lation of analogue sections, which deteriorates the overall stabil
ity. That is why the indicated values should be considered as ten
tative and are to be confirmed.
3.6 Attenuation/frequency distortion
Attenuation/frequency distortion for the whole 4wire chain
should not exceed the values given in Figure 1/G.132.
For mixed chains (without consideration of switching centre
distortions) the accumulation law of attenuation/frequency distor
tions is expressed by the following formula:
with
n1  number of analogue sections;
n2  number of analogue/digital conversions;
aF\dDM  average value (determined component) of
attenuation/frequency distortions of the analogue sections;
~F\dDM  r.m.s. deviation of attenuation/frequency
distortions of analogue sections;
aP\dC\dM  attenuation/frequency characteristics of
analogue/digital equipment;
K = 1, 2 or 3: factor defining the probability of
maximum/minimum value of attenuation/frequency distortions.
"K" is usually taken as equal to 3. The justification of the
choice for K = 3 depending on a given probability can be found in
[1, 2].
Note 1  Attenuation/frequency characteristics of
analogue/digital equipment of the same type are similar. That is
why, if in a mixed/chain of circuits analogue/digital equipment of
the same type is used, in the sum formula (1)
i =1
fIn2above ~aiPCM
can be replaced by a product n2aP\dC\dM.
Note 2  The analoguedigital equipment distortion limits
recommended in Recommendation G.712 (S 1, Figure 1) and the
FDMchannel equipment distortion limits recommended in
Recommendation G.232 (S 1, Figure 1) meet the limits indicated in
Recommendation G.132 for mixed circuits in which the number of sec
tions does not exceed 4.
When composing mixed chains with a greater number of sections,
it is advisable to utilize modern channel equipment the
attenuation/frequency distortions of which are considerably lower
than those indicated in Recommendations G.232 and G.712.
Note 3  Attenuation/frequency distortions are measured rela
tive to the reference frequency of 1020 (1000) Hz.
Note 4  See Recommendation Q.45 (S 3.4 and Q.553) to take
account of the switching equipment distortions.
3.7 Group delay distortions
Group delay distortions should not exceed the values indicated
in Recommendation G.133 for the 4wire chain.
The law of imposition of group delay distortions is expressed
by the following formula:
where
n1 the number of analogue sections,
n2 the number of analogue/digital conversions.
Note 1  If, in a mixed chain, analogue/digital equipment of
the same type is used, then the sum
i =1
fIn2above ~ ~iPCM
is substituted by a product n2  (mu  (*tP\dC\dM.
Note 2  It is expected that the group delay distortion in
mixed chains will be less than that of a fully analogue link for
any combination of analogue and digital sections. But nevertheless
the characteristics of distortion (symmetry) can change consider
ably. This should be taken into account when transmitting data on
mixed circuits containing group delay equalizers.
Note 3  Group delay distortions are measured with reference
to a frequency situated at the lower band end of the analogue chan
nel, i.e. 190200 Hz.
Note 4  Switching centre distortions are negligible and can
be ignored.
3.8 Intelligible crosstalk
Nearend and farend signaltointelligible crosstalk ratios
between circuits and between send and receive directions should
satisfy Recommendation G.151 (S 4).
Note 1  It is expected that the values indicated in
Recommendation G.151, will be maintained and even better for mixed
chains for any combination of analogue and digital sections, due to
higher values achieved in the analogue/digital conversion equip
ment.
Note 2  Measurement of the signaltocrosstalk ratio between
circuits can be performed without feeding an auxiliary signal into
a channel affected by crosstalk (unlike that provided for in the
note to point 11 of Recommendation G.712). This can be explained by
the fact that in a mixed circuit, as a rule, and in an analogue
circuit noise will be present at the input of analogue/digital con
verters in a mixed chain.
3.9 Nonlinear distortions
The existing Recommendations for analogue circuits (M.1020,
S 2.11), for switching equipment (Q.45, S 6.1) and
Recommendation G.712 for analogue/digital equipment contain dif
ferent specifications for nonlinear distortions, the methods of
their measurement differ too. The Recommendations for digital cen
tres (Q.551 to Q.554) do not contain specifications for nonlinear
distortions.
At present it is not possible to recommend permissible values
of nonlinear distortions and a method for measuring mixed chains
of circuits. This question needs to be studied.
3.10 Noise (total distortions)
The notion of noise in mixed chains of circuits due to
analoguetodigital conversions producing quantization distortions
which accompany the signal has lost its initial meaning and there
fore instead of the term "noise" applicable to mixed chain of cir
cuits the term "total distortions" is used very often. This is
stipulated by the fact that the measurement of quantization distor
tions (Recommendation Q.132) includes part of nonlinear distor
tions and singlefrequency interferences.
From this view point the total distortions in mixed chains
include analogue section noise which depends on the length of the
sections in case of terrestrial transmission systems and on the
quantization distortion which are determined by the number and type
of analoguetodigital conversions.
The addition law of total distortions is expressed by the fol
lowing formula:
P = 10 log1\d0



10DlF2619(mu FDM + 10 0.1
S  
(S/N) 10 log n2qdu






(3)
where
 WF\dDM noise power of analogue sections (pWp0)
 WF\dDM = Wo m
___ L km
(for a section provided by a satellite the terrestrial length is
taken to be equal to 2500 km).
 S/N signaltoquantization distortion
ratio of one analoguetodigital conversion.
 n2 qdu total number of quantization dis
tortion units of analoguetodigital conversions.
To determine S/N and the total number of qdu's one should
refer to Recommendation G.113.
 S signal level at which general distor
tions are measured.
To eliminate any effect of nonlinear distortion the value of
S should be no more than 10 dBm0.
The permissible value of P is to be determined in the studies
in Study Group XII.
The value of 36 dBm0 (with S = 10 dBm0),
i.e. signaltototal distortions ratio 26 dB, can be indicated as a
premilinary value.
The noise in an idle channel should comply with
Recommendations G.123 and G.153, S 1.
Note 1  Total distortions also include a component deter
mined by errors in digital sections. It is assumed that if BER at
each digital section is 10DlF2616 (with the bit rate of 64 kbit/s)
the respective component can be omitted.
Note 2  The values of total distortions for various length
of analogue sections and various numbers of qdu's mixed chains are
available in Tables 5/M.580 and 6/M.580 of Annex A to this Recom
mendation.
3.11 Single tone interference
The level of any single tone signal should not exceed 73 dBm0
(see Recommendation G.151, S 8). The indicated value does not
relate to the interfering signal at the sampling frequency.
The level of the interference at the sampling frequency should
not exceed the value of 50 + 10 log n2where n2is the number of
analogue/digital conversions in a mixed circuit. The indicated
value is tentative and needs to be confirmed by study results in
Study Group XII.
3.12 Products of unwanted modulation
Product levels of unwanted modulation caused by power sources
should not exceed 45 dB (see Recommendation G.151, S 7).
3.13 Impulse noise
Impulse noise is specified for analogue circuits used for data
transmission (Recommendations M.1020 and M.1025) and for switching
equipment (Recommendation Q.45, S 5.2 and Q.553). For
voicefrequency circuits in PCM transmission systems the impulsive
noise is not specified because it is supposed that it should not be
there at all. In practice, it has been noticed, however, that with
accumulation of errors, impulse noise can appear in a
voicefrequency circuit which leads to interference in the
transmisison of data signals. (Preliminary results on the effect of
digital link errors on impulse noise in idle PCM voicefrequency
channels is given in [4].)
The effect of impulsive noise appearing in digital sections on
the overall value of interference in a mixed 4wire chain is
subject of study.
3.14 Shorttime interruptions, phase jitter, amplitude and
phase hits
These parameters strongly influence data transmission. For
analogue circuits they are specified in Recommendations M.1020,
M.1060 and M.910. For voicefrequency circuits set up on PCM sys
tems, objectives are not available. It can be tentatively presumed
that in mixed chains of circuits the presence of digital sections
does not have a considerable effect. However, the question needs to
be studied.
3.15 Error performance
Further study.
References
[1] Moskvitin (V.  .): Opredelenije trebovanij k chastot
nym kharakteristikam zvenjev sostavnykh kanalov i traktov. (Specif
ication of requirements for attenuation frequency distortions in
sections of composite circuits and links). "Elektroviaz", 1969,
No. 11.
[2] Moskvitin (V.  .): Nozmirovanije chastotnykh kharak
teristik ostatochnogo zatuhanija kanalov. (Frequency distortion
objectives for transmission loss.) "Elektrosviaz, 1970, No. 1.
[3] COM XII19 (period 19851988), USSR
Attenuation/frequency distortions and delay distortions of mixed
audiofrequency analogue/digital circuits.
[4] COM XII188 (period 19851988), USSR Interrelation
between errors of a digital line and impulse noise in
voicefrequency channels of the PCM System.
ANNEX A
(to draft Recommendation G.136)
H.T. [T1.29]
TABLE 5/M.580
Signaltototal distortion ratio for public telephone circuit
maintenance
using a test frequency level of 10 dBm0
_________________________________________________________________________________________________________________________________________________________________________________
{
Type of circuit Number of QDUs (Note 1) Unit
< 320 321 to 640 641 to 1600 1601 to 2500 2501 to 5000 5001 to 10000 10  01 to 20  00
_________________________________________________________________________________________________________________________________________________________________________________
Analogue 0  (Note 2) dB 45 43 41 39 36 33 30
_________________________________________________________________________________________________________________________________________________________________________________
0.5 dB 35 35 34 34 33 31 29
Composite circuit 1  dB 33 33 32 32 31 30 28
_________________________________________________________________________________________________________________________________________________________________________________



















































































































Note 1  The number of QDUs contributed by various processes are
given in Table 1/G.113 [8].
Note 2  The values are idle noise terminated with a nominal
impedance of 600 ?73.
Note 3  The section of the circuit provided by satellite (between
earth stations), employing FDM techniques, contributes approxi
mately 10  00 pWp (50 dBm0p) of noise. Therefore, for the purpose
of determining the total distortion limits for international public
telephony circuits, the length of this section may be considered,
from Table 4/M.580, to be equivalent to 2500 km.
Table 5/M.580 [T1.29], p.
Blanc
H.T. [T2.29]
TABLE 6/M.580
Signaltototal distortion ratio for public telephone circuit
maintenance
using a test frequency level of 25 dBm0
_________________________________________________________________________________________________________________________________________________________________________________
{
Type of circuit Number of QDUs (Note 1) Unit
< 320 321 to 640 641 to 1600 1601 to 2500 2501 to 5000 5001 to 10000 10  01 to 20  00
_________________________________________________________________________________________________________________________________________________________________________________
Analogue 0  (Note 2) dB 30 28 26 24 21 18 15
_________________________________________________________________________________________________________________________________________________________________________________
0.5 dB 29 27 26 24 21 18 15
Composite circuit 1  dB 28 27 25 23 21 18 15
_________________________________________________________________________________________________________________________________________________________________________________



















































































































Note 1  The number of QDUs contributed by various processes are
given in Table 1/G.113 [8].
Note 2  The values are idle noise terminated with a nominal
impedance of 600 ?73.
Note 3  The section of the circuit provided by satellite (between
earth stations), employing FDM techniques, contributes approxi
mately 10  00 pWp (50 dBm0p) of noise. Therefore, for the purpose
of determining the total distortion limits for international public
telephony circuits, the length of this section may be considered,
from Table 4/M.580, to be equivalent to 2500 km.
Table 6/M.580 [T2.29], p.
ANNEX B
(to draft Recommendation G.136)
SOURCE: THE URSS TELECOMMUNICATION ADMINISTRATION
TITLE: INTERRELATION BETWEEN ERRORS IN A DIGITAL
CIRCUIT AND IMPULSE NOISE IN VOICEFREQUENCY CHANNELS OF THE PCM
SYSTEM
B.1 Introduction
Voicefrequency channels of PCM as well as FDM systems should
be fit for transmitting various types of signals. It is well known
that the transmission quality of discrete signals in
voicefrequency channels is affected by impulse noise. At present,
Recommendation G.712 has no requirements to voicefrequency
PCMchannels regarding impulse noise. However, under reallife con
ditions in a voicefrequency PCM channel impulse noise contributes
to the errorrate of digital links. The present contribution gives
the investigation results of impulse noise in voicefrequency
PCMchannels.
B.2 Influence of digital circuit errors on impulse noise in
an idle voicefrequency PCM channel
Evaluation of error influence on digital links on the value of
impulse noise in voicefrequency channels was conducted experimen
tally on a channel equipment (satisfying Recommendation G.712) of a
PCM transmission system (2048 kbit/s). With the help of an error
simulator errors had been introduced into one or several bits
corresponding to a chosen idle voicefrequency channel of a digital
link (Figure 1). In the voicefrequency channel impulse noise could
be observed with the help of an oscillograph. The shape of the
pulse response in the voicefrequency channel is presented in
Figure B2.
Figure B1, p.
Figure B2, p.
The parameters of pulse response are given in Table 1 (the
values are chosen for the point of the relative zero level at a
resistance of 600 ohms). These data allow us to formulate the fol
lowing conclusions:
 The pulse amplitude of the response depends on
the bit number which contains the error; the errors in the more
significant bits cause a greater amplitude of the response.
 With single errors the maximum value of the pulse
peak A1(in case of an error in the second bit) is 22.1 dBm0.
 With burstbuilding and with an increase in the
number of errored bits in the code word of the prime digital path
(2048 kbit/s) the response amplitude values A1, A2, A3, .  
grow, but their duration, as determined by the response of the
channel's low frequency receiving filter, remains unchanged. This
applies to the cases where in a prime digital path, the error
bursts affect the digital stream for not more than one discretiza
tion period, i.e. the number of the errors in a burst does not
exceed 256. With errors in code words occurring every 125 us the
superposition of responses takes place as a result of the receiving
filter reaction on the error pulses in each following discretiza
tion period.
H.T. [T3.29]
TABLE B1
_________________________________________________________________________
{
t 1 t 2 t 3
_________________________________________________________________________
2 22.1 28.2 33.8 320 160 130
3 34.1 40.2 45.8 320 160 130
2 and 3 10.1 16.2 21.8 320 160 130
2 and 3 and 4 from 2 to 8, 4.1 10.2 15.8 320 160 130
{
2 discretization periods
from 2 to 8,
} +4.3 6.7 14.8 440 180 100
{
3 discretization periods
from 2 to 8,
} +4.3 4.9 14.8 600 200 100
{
4 discretization periods
from 2 to 8,
} +4.3 4.7 14.8 680 180 120
{
5 discretization periods
from 2 to 8,
} +4.3 6.7 14.8 840 200 120
{
6 discretization periods
from 2 to 8,
} +3.8 4.3 14.8 930 200 100
7 discretization periods +5.25 8.7 14.8 1100 180 140
_________________________________________________________________________




































































































































































































































































Table B1 [T3.29], p.
Thus, when errors, on a 2048 kbit/s digital path grow into
burst of 2 errors and more there is a certain probability that the
value of the impulse noise in a PCM voicefrequency channel exceeds
21 dBm0 given in Recommendation M.1020, S 2.6.
With error bursts of 256 and more bits the abovementioned
impulse noise will always be present.
The quantitative relationship between the number of bursts,
the number of errors in them within a definite time interval and
the number of impulse noise interferences and the BER in a
voicefrequency channel is under study at present.
Supplement No. 30
TRANSMISSION PLAN ASPECTS OF LAND MOBILE TELEPHONY NETWORKS
Draft Recommendation G.173
(This Supplement is proposed for study during the present
study period with the aim to convert it into a Recommendation.)
1 General
This Recommendation is primarily concerned with the special
planning aspects which pertain to analogue or digital land mobile
systems. Such systems, due to technical or economic factors, will
prevent a full compliance with the general characteristics of
international telephone connections and circuits recommended by
CCITT.
The scope of this Recommendation is thus to give guidelines
and advice to Administrations as to what kind of precautions, meas
ures and minimum requirements which are needed for a successful
incorporation of such networks in the national PSTN.
The performance objectives of such systems may vary between
different groups of customers. For normal customers the objective
should be to reach a quality as close as possible to CCITT stan
dards. For other groups of very disciplines customers other perfor
mance objectives might be acceptable.
2 Network configurations
Under study.
Under this headline Administrations should be advised to use
4wire transmission to avoid problems when accessing inherently
4wire mobile links.
3 Nominal transmission loss of mobile links
Under study.
Under this headline the problems with the application of loud
ness ratings and the correct loading of the radio channels should
be discussed.
The recommended LR values in CCITT Recommendation G.121 are
not directly applicable due to the fact that the background noise
level is higher in a car than what is assumed in
Recommendation G.121.
What is the design objective for the speech levels from the
radio path and what levels should be delivered to the network?
4 Stability
Under study.
5 Echo
Under study.
Under this headline the need for echo control devices should
be discussed.
6 Noise
Under study.
(Can the European group give indications of the inherent noise
performance of the codec algorithms being considered?)
7 Delay
Under study.
8 Effects of errors in digital systems
Several coding methods, such as SBC, ATC, RELP and APCAB with
transmission bit rates 16 kbit/s have been proposed to achieve
spectrum utilization efficiency and quality comparable with conven
tional analog FM systems. However, the application of such highly
efficient speech coding methods to land mobile radio can lead to a
significant degradation in quality because of transmission errors.
Mobile radio links are not always errorfree. Burst errors
occur frequently due to multipath fading. It has been reported that
the average bit error rate (BER) performance of diversity reception
is 10DlF261210DlF2614 in the 10 to 20 dB range of the average car
rier to noise power ratio (CNR), and burst error length reaches
20 to 100 bits in case of 16 kbit/s digital signal transmissions.
Therefore, robustness against burst error is an important charac
teristic for speech coding applied to mobile communication. Speech
CODECs in mobile radio links should involve error control tech
niques so as to provide robustness in multipath fading channels.
Thus, the transmission bit rate includes redundancy bits for error
control.
Concerning quality evaluations, it may be better to use the
average CNR as the receiving level for comparisons among analogue
and digital systems. This is because it can present the receiving
level as a normalized unit for both analog FM and digital systems.
In quality evaluations between digital systems, the average signal
energy per bit to noise power density ratio (Eb/No) is suitable for
the presentation of the receiving level. This is because it can
describe the receiving level as a normalized unit for any transmis
sion bit rate and receiving bandwidth.
9 Quantizing distortion
Under study.
10 Effect of transmission impairments on voiceband data performance
Under study.
Blanc