5i'
PART II
SUPPLEMENTS TO RECOMMENDATIONS IN
SECTIONS 2 TO 5 OF THE SERIES G RECOMMENDATIONS
MONTAGE: PAGE 204 = PAGE BLANCHE
Supplement No. 4
CERTAIN METHODS OF AVOIDING THE TRANSMISSION
OF EXCESSIVE NOISE BETWEEN INTERCONNECTED SYSTEMS
(referred to in Recommendation G.221;
for this Supplement see page 572, Volume III of the Green Book, Geneva,
1973)
Supplement No. 5
MEASUREMENT OF THE LOAD OF TELEPHONE CIRCUITS  fR UNDER FIELD
CONDITIONS
(referred to in Recommendations G.223 and H.51;
for this Supplement see page 295, Fascicle III.2 of the Red Book,
Geneva, 1985)
Supplement No. 6
EXAMPLE SHOWING HOW THE TOTAL VALUE OF LINE NOISE
SPECIFIED FOR THE HYPOTHETICAL REFERENCE CIRCUIT
ON OPENWIRE LINES MIGHT BE BROKEN DOWN
INTO ITS VARIOUS COMPONENTS
(referred to in Recommendations G.223 and G.311;
for this Supplement see page 589, Volume III of the Green Book, Geneva,
1973)
Supplement No. 7
LOSSFREQUENCY RESPONSE OF CHANNELTRANSLATING EQUIPMENT
USED IN SOME COUNTRIES FOR INTERNATIONAL CIRCUITS
(referred to in Recommendation G.232;
for this Supplement see page 590, Volume III of the Green Book, Geneva,
1973)
Supplement No. 8
METHOD PROPOSED BY THE BELGIAN TELEPHONE ADMINISTRATION
FOR INTERCONNECTION BETWEEN COAXIAL AND SYMMETRIC PAIR SYSTEMS
(referred to in Recommendation G.322;
for this Supplement see page 591, Volume III of the Green Book, Geneva,
1973)
Supplement No. 9
ROLL EFFECT IN COAXIAL PAIR SYSTEMS
(referred to in Recommendation G.333;
for this Supplement see page 592, Volume III of the Green Book, Geneva,
1973)
Supplement No. 13
NOISE AT THE TERMINALS OF THE BATTERY SUPPLY
(referred to in Recommendation G.229;
for this Supplement see page 664, Volume III.3 of the Orange Book,
Geneva, 1977)
Supplement No. 17
GROUPDELAY DISTORTION PERFORMANCE OF TERMINAL EQUIPMENT
(referred to in Recommendations G.233 and G.242;
for this Supplement see page 311, Fascicle III.2 of the Red Book,
Geneva, 1985)
Supplement No. 22
MATHEMATICAL MODELS OF MULTIPLEX SIGNALS
(referred to in Recommendation G.223;
for this Supplement see page 326, Fascicle III.2 of the Red Book,
Geneva, 1985)
Supplement No. 23
EXPLANATORY NOTES FOR THE INFORMATION OF DESIGNERS
OF A MARITIME MOBILE SATELLITE SYSTEM
(Geneva, 1980; referred to in Recommendation G.473)
1 Allocation of losses in the maritime system
1.1 Complying with Recommendations
1.1.1 Figure 1 illustrates the nomenclature adopted in this
Supplement and the arrangements for a 2wire switched shipboard
installation.
Figure 1 p.
1.1.2 The CCITT Recommendations which jointly influence the
choice of the losses in the path (a  b ) and the reference
equivalents of the local system referred to an appropriate set of
terrestrial virtual switching points are as follows:
Recommendation G.122 [3]  To ensure that international
connections have an adequate stability the loss (a  b ) at any
frequency in the band 0 to 4 kHz should not be less than (6 + n
) dB where n is the number of 4wire circuits in the national
chain.
Recommendation G.131 [4]  When calculating stability the
variations of loss in the two directions of direction are taken to
be fully correlated.
_________________________
Note of the Secretariat  A revision of Recommenda
tions G.111 [1] and G.121 [2] has been adopted intro
ducing the new concept of corrected reference
equivalents. The values of reference equivalents are
maintained here for the next Study Period to give
planners sufficient time to become acquainted with the
new concepts.
Recommendation G.151 [5]  The standard deviation of
transmission loss for modern national and international circuits
should not exceed 1 dB.
We seek a primafacie assurance that the contribution of the
maritime extension to the stability of the 4wire chain is not
worse than that of a comparable national extension. The factors
affecting the stability are mean departure from nominal, standard
deviation of loss, and attenuation distortion. The mean departure
from nominal and standard deviation are expected to be about twice
the corresponding values for a terrestrial circuit so that the one
satellite circuit can be regarded as having the same effect as four
terrestrial FDM circuits when full correlation is assumed. As far
as attenuation distortion is concerned, since the channel equip
ments in the shore station are not in permanent association with
those on board ship, betweenchannel variations manifest themselves
as another source of variance among the connections and an
allowance of 1 dB is made for this effect.
The formula in the Recommendation cited in [3] can be rewrit
ten as (6 + 1n ) in which the coefficient 1 dB/circuit is explicit
rather than implicit, and we have derived the
coefficient 4 + 1 = 5 dB/circuit for the case of the satellite cir
cuit. Hence, with n = 1 we obtain the condition:
S + R + B _" 11
Recommendation G.161, Test 8 [6]  The equivalent level
go/return loss on the officeside of the echo suppressor should not
be less than 6 dB. In principle, this quantity, which can be
derived from the relative levels at the 2wire switch point, is to
be evaluated under conversational conditions at any frequency
within the detection band of the echo suppressor.
Recommendation G.121 [2]  The various constraints on the
reference equivalents and losses of national systems are as fol
lows, in which the overbar indicates an average value:
SRE:  fR preferred range: 10 to 13 dB
permissible range: 10 to 16 dB
RRE:  fR preferred range: 2.5 to 4.5 dB
permissible range: 2.5 to 6.5 dB
i.e.: 10 S + s  fR 13 or 16; 2.5 R + r  fR 4.5
or 6.5
These values obviously take into account variations
although we shall assume that the variability of s and r are small,
i.e. that s  fR = s and r  fR = r .
SRE max = 21 dB, i.e.: S + s 21
RRE max = 12 dB, i.e.: R + r 12
These are 97% planning values but we shall take them as
100% planning values.
SRE min = 6 dB, i.e.: S + s _" 6
Ideally this should take variations into account but the
recommendation is in terms of a planning value.
Difference between the losses (a  t ) and (t  b )
should not exceed 4 dB, i.e.:  fIS  R  4.
1.1.3 The Recommendations do not enable us to ascertain the
separate values of S and R because the CCITT does not specify any
particular subdivision of national reference equivalents as between
the local system and the circuits in the remainder of a national
extension.
We may simplify the problem by assuming that during the setup
or cleardown of the connection it is not possible to prevent B
= 0, so that S + R = 11 and furthermore we shall aim to stay
within the preferred ranges of mean values of the reference
equivalents recommended for a national system.
1.1.4 It is clear that within the constraints of:
S + R = 11;  fIS  R  4;  fIS

S  fR 
0.5;  fIR 
R  fR  0.5; 10
SRE  fR 13; 2.5
RRE  fR 4.5
the individual values of S and R can be chosen to permit a range of
reference equivalents for the shipboard local system as illustrated
by the solution domains shown in Figure 2. We must therefore seek
other criteria on which to base a decision.
Figure 2 p.
1.2 Subjective criteria
1.2.1 Figure 3, which is based on the corresponding one in
Recommendation G.473, illustrates various minimum, average, and
maximum configurations utilizing the information concerning traffic
routing contained in [7]. These routings have been used to develop
hypothetical reference connections based on
Recommendation G.103 [8] to enable the effects of loss, noise, and
distortion to be studied in accordance with the principles outlined
in the CCITT manual cited in [9].
Figure 3 p.
1.2.2 A few S, R and hence s, r values within the permitted
solution domains have been studied in order to determine an optimum
set from the point of view of subscriber opinion. The results of
two such calculations are recorded in Table 1.
In one calculation the S and R values were equal (i.e.: the
S/R differential was zero) and the SRE  fR and RRE  fR values
were in the middle of their preferred ranges ( SRE  fR = 11.5;
RRE  fR = 3.5; S = 5.5; R = s = 6; r = 2). In the other cal
culation half the permitted S/R differential was introduced reduc
ing the RRE  fR at the expense of the SRE  fR but nevertheless
keeping their values within 0.5 dB of the extrema of their pre
ferred ranges to allow for the 0.5 dB mean departure from nominal
in the values of S and R . ( SRE  fR = 12.5; RRE  fR = 3; S
= 6.5; R = 4.5; s = 6; r = 1.5).
Table 1 T1.23, p.
1.2.3 Table 1 shows that moving from the centre of the pre
ferred ranges has hardly any effect on the opinion scores of the
shipboard customer but has a somewhat greater, worsening effect on
the inland customer particularly on the maximum routing. Hence we
advocate arrangement B in which the SRE  fR and RRE  fR are at
the middle of their preferred ranges. This arrangement also has the
advantage that the inland and shipboard customers' opinions are
more nearly equal in the case of the average routing (which we must
assume will carry the most traffic). This is only true for %D , not
for %P + B
1.2.4 Figure 4 illustrates how opinion worsens as the design
noise power of the maritime satellite system increases from 10 
00 pW0p (50 dBm0p) to 100  00 pW0p (40 dBm0p). These are the
effective design noise power levels, i.e.: either an uncompandored
noise power level, or the result of a 2:1 compandor with a 0 dBm0
unaffected level operating on a higher noise power level, in which
case the empirical rule onethird speechon noise power level plus
twothirds speechoff noisepower level was used to calculate the
effective noise (see [11]).
Note  The reference equivalents of the maritime system are
SRE  fR = 11.5 and RRE  fR = 3.5. Only the effects of loss,
noise, bandwidth limitation, and attenuation/distortion have been
estimated. The effects of delay and imperfectlysuppressed echo
could not be taken account of in the calculations but they should
not be disregarded.
Figure 4 p.
1.3 The preferred arrangement with 2wire switching
1.3.1 Figure 5 illustrates the preferred arrangements upon
which Recommendation G.473 is based. All the arrangements are in
terms of the CCITT quantities which relate to virtual switching
points on the international circuit.
It is clear that SRE min _" 6 is comfortably met in these
arrangements, using planning values. We note that the limit is also
met even when variation is allowed for; the 2.33 ~ value for the
midrange value of the SRE is 11.5  0.5  2.33(2) = 6.3 (rounding
down to the nearest 0.1).
1.3.2 Diagram I of Figure 6 gives an example of how
Arrangement I could be realized in a practical installation. We
have assumed the following:
 actual 4wire switching levels of 2 dBr, which
is typical of many international switching centres;
 a 2wire sending level of 0 dBr, which is suit
able for a local system with a nominal midrange value of sending
reference equivalent of 6 dB (SRE ) connected to that point;
 symmetrical 3.5 dB terminating units;
 channel translating equipments operated at rela
tive levels of +4 dBr/14 dBr, a pair of levels appearing in
Recommendation G.232;
 farend operated, halfecho suppressors introduc
ing 0 dB transmission loss at the appropriate relative levels.
1.3.3 It remains to calculate the equallevel go/return loss
on the office side of the echo suppressor which is seen to be
9.5 + 3.5 + B + 3.5 + 10.5  18 = 9 + B
thus complying with the test conditions of
Recommendation G.161 [12] (assuming no negative Bvalues). The
quantity 10 can be obtained with more insight and less arithmetic
by taking the difference of the two relative levels at the 2wire
switch point.
Figure 5 p.
Figure 6 p.
1.4 Arrangements with 4wire switching
1.4.1 Figure 5 also illustrates two other arrangements which
incorporate 4wire switching: Arrangement II which retains a 2wire
handset, and Arrangement III which is wholly 4wire. Examples (for
guidance only) of the corresponding practicable realizations are
shown in Figure 6.
1.4.2 A halfecho suppressor is shown in Figure 6 for the
wholly 4wire case. This is to control, if necessary, the echo that
might arise from the acoustical path via the handset of the ship
board subscriber. The total echo loss demanded between virtual
switching points is effectively 56 dB (a consequence of the Recom
mendations cited in [15] and [3]). The minimum (electrical)
go/return loss is required to be of the same order (the Recommenda
tion cited in [16]) so as not to nullify the effect. It is clear
that the shipboard installation should aim at a comparable perfor
mance. The recommended values of RRE and SRE shown in Figure 5 add
up to 15 dB which implies that the acoustic go/return loss must
fall below 41 dB. As the
system designer has no control over how the ordinary subscriber
uses his handset there is a prima facie case for assuming little
possibility of being able to guarantee this value. However, there
is little experimental data on this topic and further study is
desirable. The results of such a study may indicate that suppres
sors can be dispensed with in wholly 4wire installations.
Fourwire head and breast sets (or presstospeak handsets) used
for special purposes by trained persons would be less troublesome
in this regard and it is unlikely that a shipboard suppressor
would be needed for these cases.
1.5 Taking advantage of a nonzero stability balance return
loss
1.5.1 All the allocations of loss shown for a 2wire local
system tacitly assume that during setup or cleardown there is the
possibility of zero balance return loss at the 2wire/4wire ter
minating set. However, if special arrangements are made, as indi
cated for example in Recommendation Q.32 [14] so that at all times
a certain minimum positive value is assured, there can be
corresponding reductions of the S , R values and increases of
the s , r values.
1.5.2 The arrangements of Recommendation Q.32 [14] introduce a
minimum of 6 dB balance return loss and assuming this to be less
than the offhook balance return loss of the shipboard local sys
tem, the S and R values could be reduced by 3 dB each and the s ,
r values correspondingly increased. Other partitions are possible,
provided they comply with constraints given in S 1.2 above. It is
clear that if S and R can be reduced in this way there is more
scope for catering for a range of existing shipboard local systems.
2 Estimated speech power levels and signaltonoise ratios
2.1 Speech power levels entering the maritime system at the
shore station
2.1.1 We can estimate the mean and standard deviation of the
speech power levels at the shore station by considering the
relevant Recommendations. Naturally this is not claimed to be the
same as measured values, but it is probably the best we can do for
planning purposes, particularly since trafficweighted values are
not really appropriate for equipment design if a worldwide service
is planned for.
2.1.2 Recommendation G.121 [2]: National systems
Mean SRE calculated to international virtual analogue switch
ing points: 13 dB
Range is (21  6) = 15 dB from which, as an approximation,
standard deviation = 1/4 (range) = 3.8 dB.
2.1.3 Recommendation P.16 [17]: Crosstalk
Conversational speech power level from an active median talker
via a 0 dB SRE end is 6 dBm; standard deviation is 4.8 dB.
2.1.4 Annex 2, Question 1/XVI [7]: International 4wire
chain
From the statistics of the international 4wire chain recorded
in [7] we obtain the estimate of the mean and the variance of the
transmission loss of this portion of the connection shown in
Table 2, assuming that the circuits comply with the provisions of
Recommendation G.151 [18] in respect of standard deviation.
Table 2 T2.23, p.
2.1.5 Combining all these estimates, the distribution of
speech power levels at the input to the maritime system at the
shore station we obtain:
Mean = 1360.6 = 19.6 dBm
Standard deviation =
\
____________________
.8 2 + 4.8 2 + 1.108 = 6.2 dB
2.1.6 We can reasonably assume 3.5 dBr to be the relative
level at the input to the maritime system directly connected to the
receive virtual switching point of the international circuit
delivering the signal, although strictly speaking there is no
recommendation concerning the relative level at these points on the
"national" side of the virtual switching points.
2.1.7 Hence we finally obtain as a defensible system planning
value:
Mean = Median = 16.1 dBm0
Standard Deviation = 6.2 dB.
2.2 Speech power level at the input to the maritime system
from the shipboard local system
In any studies concerning a fixed threshold setting (e.g. for
echo suppressor or noise squelch circuit) it should be noted that
the midrange value of the sending reference equivalent referred to
a 0 dBr point for the shipboard local systems illustrated in Fig
ure 6 is 6 dB corresponding to a mean active speech power level of
11.5 dBm0, so that 99% of talkers would not fall below 12 
2.33(4.8) = 23.5 dBm0. This would thus be a suitable level for a
threshold detector based on mean active power level. A detector
responding to syllabic power levels would need to be set somewhat
lower if centre clipping effects are to be avoided. If an increase
of the s value is foreseen (as a result of the considerations out
lined in S 1.5.2 above) the mean active speech power level will be
correspondingly reduced.
2.3 Distribution of speech signaltonoise ratios at the
output of the maritime system on board ship
2.3.1 What follows is an elementary estimate of the distribu
tion of the speech signaltonoise ratio in a switched telephone
network, in which there is a distribution of speech volumes, using
a maritime satellite system achieving the longterm aim of a design
noise power level of 50 dBm0p (10  00 pW0p) regarded as essen
tially constant for most of the time. This, of course, represents a
reversal of the basis on which conventional HF radio circuits are
designed in which the speech volume is assumed to be held substan
tially constant by means of a constant volume amplifier (or a
technical operator), and the noise being regarded as the variable.
2.3.2 Signal  Since the distribution of speech volumes is
substantially lognormal, the speech power level of the active
median talker is given by:
10 log1\d0(mean power/1 mW)  0.115 ~2
where ~2 is the variance of the distribution of speech power lev
els. Allowing, say 2 microwatts for echos and other currents, the
conventional load for the speech power at a 0 dBr point averaged
over all channels is 20 microwatts and the conventional activity
factor is 0.25. Hence the (conventional) mean active speech power
is 80 microwatts. The standard deviation of speech volumes is of
the order of 6.2 dB (see S 2.1.5 above). We obtain from these fig
ures:
speech power level of the active median talker
= 10 log1\d0(80/1000)  0.115(38.44) = 15.4 dBm0
Noise  In the case being considered, i.e. the longterm aim,
the constant equivalent value is 50 dBm0p.
2.3.3 Hence, the mean signaltonoise ratio Q  fR , is Q  fR
= S  fR  N  fR = 15.4 (50) = 34.6 dB. Q is normally dis
tributed with a standard deviation of 6.2 dB, and the principal
source of variation in the signal level will arise either from dif
ferent talkers on the various channels provided by the maritime
satellite link, or from successive talkers on a particular channel,
i.e.: it is assumed that the process is essentially ergodic. Hence
we can construct Table 3 showing the various percentages of time
(to the nearest 1%) for which particular values of signaltonoise
ratios are exceeded by setting k = (Q  34.6)/6.2 and consulting
tables of the normal variate.
Table 3 T3.23, p.
2.3.4 In the case of the shortterm limits for noise,
Figure 4/G.473 defines the following functional relationships
between S (speech signal) and Q (signal/noise ratio) as
[Formula Deleted]