Telecommunications Industry Association (TIA)
Clearwater Beach, FL
TR-30.3/99-12-111
December 1, 1999
COMMITTEE CONTRIBUTION
Technical Committee TR-30.3
SOURCE:
Lucent Technologies
CONTRIBUTOR:
Horace Hagen
732-949-3586
732-949-6868 (FAX)
hhagen@lucent.com
TITLE:
Proposed International Round Trip Delays
PROJECT:
PN-3857
DISTRIBUTION:
Members of TR-30.3 and meeting attendees
ABSTRACT
This document presents the range of round trip delay (RTD) values that need to be included in an
international version of PN-3857. The RTDs are derived from a network model based on intercontinental
submarine cable and geostationary satellite facilities in-service in 1995. Calculated cable and satellite
delays were combined with measured voice compression (DCME) terminal delays and estimated tail
circuit delays. The model provides an estimated worldwide end-to-end RTD range for submarine cable
connections of 62 - 500ms and an estimated end-to-end RTD range for satellite connections of 510 880ms. The RTDs measured during international modem tests provided a sanity check on the delay
model. Comparisons between the estimated delays and modem measurements indicate good
agreement. The RTD ranges for submarine cable and satellite connections decrease to 62 – 280ms and
510 – 660ms, respectively, if the DCME is removed.
Copyright Statement
The contributor grants a free, irrevocable license to the Telecommunications Industry Association (TIA) to incorporate text contained in this
contribution and any modifications thereof in the creation of a TIA standards publication; to copyright in name any standards publication even
though it may include portions of this contribution; and at TIA’s sole discretion to permit others to reproduce in whole or in part the resulting TIA
standards publication.
Intellectual Property Statement
The individual preparing this contribution does not know of patents, the use of which may be essential to a standard resulting in whole or in part
from this contribution.
2
INTRODUCTION
Transmission delays in the international switched telephone network have increased due to the
introduction of fiber optic cables, geostationary satellite transmission facilities and digital circuit
multiplication equipment (DCME). Delays decreased to some countries when cable access was
substituted for exclusive satellite access. The transmission delay of fiber optic cable is approximately 25
percent greater than that of coaxial cable. The largest increase in delay occurred when voice
compression equipment (DCME) was installed on essentially all intercontinental undersea cable and
geostationary satellite facilities during the mid eighties. The DCME introduces one way processing
delays ranging from 26 to 110 milliseconds (ms). The increases in delay affect high speed (echo
canceler based) modem designs by requiring longer bulk delay lines (more memory) to accommodate the
Round-Trip Delays (RTDs) associated with far (talker) echoes. Reference is made to RTDs measured
during an international modem test program run in 1993. These data provide a sanity check on RTDs
generated by the described network RTD model.
MODEM DATA BASE
A large number of international test calls (over 800 calls to 18 foreign countries) were made with a V.34
modem. The calls were made during 1993 and originated from New Jersey and Indiana. The modem
provided diagnostic information including the RTD with an accuracy of 10 milliseconds.
A histogram of the RTDs associated with the calls routed over terrestrial or submarine cable facilities is
displayed in Figure 1. The delays ranged from 80 to 420ms and the average delay was calculated to be
170ms. In plotting the data, the maximum expected terrestrial cable delay was limited to 450ms. The
surprisingly long terrestrial delays ranging from 380 to 420ms were all encountered on calls to the
Philippines. The exact source of the long delays is not known but the delays were most likely due to
indirect call routing or large processing delays associated with a specific DCME type.
A histogram of RTDs associated with calls routed over satellite facilities is shown in Figure 2. The delays
ranged from 480 to 820ms and the average RTD was 620ms. A few test calls were forced over facilities
without DCME. The effect of DCME on the delays is made visible by the peaks at 590 and 640ms
(590ms=uncompressed and 640ms=compressed). These peaks verify that some of the DCME
equipment was the ECI DTX-240 because it exhibits a RTD of 52ms. The single data points at 480 and
500ms were measured on calls to Japan and Poland, respectively. A delay of 480ms to Japan is
physically impossible on a satellite connection because an earth station on the East Coast of the USA
does not have access to a satellite over the Pacific Ocean. Thus, a satellite call must first traverse the
USA terrestrially (60ms) and then continue on a satellite from the West Coast to Japan resulting in a
minimum total RTD of about 580ms. The 480ms delay is probably due to an indirect cable route, DCME
processing delay or a database error. The 500ms delay to Poland cannot be explained and is assumed
to be a database error. Delays greater than 750ms were encountered on calls to Alaska, India, Italy and
Poland because the associated DCME was programmed for a processing delay of up to 220ms. One call
to India connected via double satellite hop and exhibited the longest RTD of 1020ms. With this number
removed the worst case RTD measured was 820ms.
DELAY ALLOCATIONS
The following sections quantify the round trip delays introduced by satellite and cable transmission
facilities and Digital Circuit Multiplication Equipment. The RTDs will be used to construct a network delay
model.
SATELLITES
The significant transmission delay introduced by a geostationary satellite adversely affects voice
communications. It is for this reason that most international carriers do not use satellite facilities for the
3
first choice trunk groups for voice calls. Because long delays affect FAX and DATA (modem) traffic to a
lesser extent, some carriers route FAX and DATA calls over satellites. However, during peak traffic
periods, satellite circuits come into play even for voice calls. Since the maximum viewing angle of these
satellites, in latitude and longitude, is approximately 130 degrees, a minimum of three satellites stationed
120 degrees apart and above the equator are required to cover most of the earth’s surface (locations at
north and south latitudes greater than 65 degrees do not have geostationary satellite access). In 1995,
there were three groups of geostationary satellites in operation. One group of four satellites was located
at 57, 60, 63 and 66 degrees over the Indian Ocean. The second group of four was located at 174, 177,
180 and 183 degrees over the Pacific Ocean and the third group of eight satellites was located at 307,
325, 332, 335, 338, 342, 346 and 359 degrees over the Atlantic ocean. With a 130-degree longitude
viewing angle limitation the satellites over the Pacific Ocean can only be accessed from the West Coast
of North America, Australia and Pacific rim countries. The range of RTDs associated with a satellite alone
is 480 to 540ms depending on latitude and longitude of the two earth stations. The manner in which
satellites and undersea cables are combined is country and carrier specific.
FIBER OPTIC CABLES
Before fiber optic cables and geostationary satellites were introduced, the dominant transmission media
utilized in the international switched telecommunications network was undersea coaxial cable. With
coaxial cable the average RTD for connections from the east coast of the USA to Europe was 55ms.
When fiber optic cable was introduced the RTD to Europe increased to 68ms. The RTD for a fiber optic
cable can be approximated by multiplying the length in kilometers by 10 microseconds per kilometer.
DIGITAL CIRCUIT MULTIPLICATION EQUIPMENT (DCME)
There are four major DCME deployed on intercontinental cables and satellites. The country of origin,
manufacturer, model, round trip delay and estimated 1995 percent deployment are listed in the following
table.
Country
Israel
France
USA
Japan
TABLE I – DCME RTDs
Manufacturer
Model
RTD
ECI
DTX-240
52ms
ALCATEL
CELTEC-3G
60ms
AT&T
IACS
50-220ms
MITSUBISHI
DX-3000
60ms
Percent
85
9
5
1
NETWORK DELAY MODEL
A Network Delay Model was developed to provide an estimate of RTD to eighteen transoceanic countries.
The foreign destinations represent countries with significant data traffic and the worst cases (half way
around the globe). The following assumptions were made in constructing the model:
1. Calls are originated from the east coast of the USA.
2. USA and foreign tail circuits are independent variables.
3. All USA based international carriers share the same cable and satellite facilities.
4. The number of calls made to a country is directly proportional to the available circuit capacity.
5. DCME is present on most intercontinental calls and the breakdown of DCME for all international
carriers is 85, 9, 5 and 1 percent for ECI, ALCATEL, AT&T and MITSUBISHI, respectively.
The RTDs for 52 undersea cable combinations (23 individual cables) to 18 transoceanic countries were
calculated. The delays are plotted in attached Figure 4 and are listed in attached TABLE A. Note that the
number of cable routes varies from 1 to 6 and is country specific. The data in Figure 4 were combined
with the delays due to various DCME and USA and foreign tail circuits. The estimated 5th, 50th, 95th and
99th percentile RTDs for the USA tail circuit, DCME and foreign tail circuit are listed in TABLE II below.
4
TABLE II - PERCENTILE RTD ADDERS IN MILLISECONDS
PERCENTILES
5%
50%
95%
USA TAIL CIRCUIT
5
30
60
DCME
52
52
120
OVERSEAS TAIL CIRCUIT
5
30
40
99%
60
180
40
The numbers were added to the calculated cable and satellite delays and resulted in the five curves in
Figure 5. Figure 5 indicates an end-to-end maximum RTD of 460ms and 800ms for terrestrial and
satellite connections, respectively. In the past, some countries were connected to the outside world
exclusively via satellite. This has changed within the last few years as additional fiber optic cables were
installed. The percentage of calls routed over satellite circuits is country specific. For the selected 18
countries used in the model the percentage of traffic expected to be routed over satellites ranged from 4.7
to 100.0 percent.
TABLE III lists the minimum and maximum (0 and 100% worst case) end-to-end RTDs that can occur
when all the component delays are at their minimum or 100% WC. For example: Calls placed from the
USA to a Pacific rim country via a 100 percent worst case USA tail circuit, SATELLITE, DCME and
foreign tail circuit can encounter a maximum RTD of 60+540+220+60 = 880ms
TABLE III - END-TO-END DELAY LIMITS IN MILLISECONDS
ATLANTIC
PACIFIC
MINIMUM MAXIMUM MINIMUM MAXIMUM
USA TAIL CKT
0
60
0
60
OVERSEAS TAIL CKT
0
60
0
60
DCME
52
220
52
220
CABLE ONLY
62
100
89
180
SATELLITE ONLY
510
540
510
540
END-TO-END CABLE
END-TO-END SATELLITE
62
510
420
880
89
510
500
880
CONCLUSION
A network round-trip delay model was constructed based on cable and satellite facility information. The
RTDs corresponding to the various cable routes were calculated and combined with estimated tail circuit
and measured DCME terminal delays. End-to-end connections were constructed using the calculated
and estimated delays. The model indicates a worldwide end-to-end RTD range for submarine cable of 62
to 500 milliseconds with an estimated average of 208ms (30+52+96+30). The RTD range for satellite
connections was determined to be 510 to 880ms with an estimated average of 632ms (30+52+520+30).
Since the delay due to the satellite alone falls between 510 and 540ms, the end-to-end delay variation is
dominated by the type of DCME and tail circuit lengths. The modem database indicated a terrestrial RTD
range of 80 to 420ms with an average of 170ms. The measured satellite RTDs ranged from 540 to
820ms with an average of 620ms. The higher average delays predicted by the model are due to the
larger number of foreign destinations. For the 18 selected foreign destinations, the network model
predicts a 100% worst case RTD of 880ms. It is important to note that the switched network is
continuously changing. Of the 23 fiber optic undersea cables studied, 9 were placed into service since
the 1993 modem tests. Additional fiber optic cables are being installed every year. This effort will
eventually result in cable connectivity for every country in the world. As the band width of these cables
increases the need for DCME will decrease and result in significant reductions in round trip delays. If all
5
DCME is removed, the new RTD range for submarine cable connections drops to 62 - 280ms with an
estimated average of 156ms. The new RTD range for satellite connections would drop to 510 - 660ms
with an estimated average of 580ms.
Attachments:
Figures 1 through 4
TABLE A
ROUND TRIP DELAY - MILLISECONDS
450
440
430
420
410
400
390
380
370
360
350
340
330
320
310
300
290
280
270
260
250
240
230
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
NUMBER OF CALLS WITH DELAY = ABSCISSA
6
FIGURE 1 - SUBMARINE CABLE CALLS (523)
50
45
40
35
30
25
20
15
10
5
0
ROUND TRIP DELAY - MILLISECONDS
850
840
830
820
810
800
790
780
770
760
750
740
730
720
710
700
690
680
670
660
650
640
630
620
610
600
590
580
570
560
550
540
530
520
510
500
490
480
470
460
450
NUMBER OF CALLS WITH DELAY = ABSCISSA
7
FIGURE 2 - SATELLITE CALLS (335)
50
45
40
35
30
25
20
15
10
5
0
8
FIGURE 3 - SUBMARINE CABLE ROUND-TRIP DELAYS
200
ROUND-TRIP DELAY IN MILLISECONDS
180
160
140
120
100
80
60
40
20
0
DESTINATION - CARRIER
9
FIGURE 4 - END TO END ROUND-TRIP DELAYS
900
800
C A BLE / SA TELLITE O NLY
PLUS 5% DC M E & TA ILS
ROUND-TRIP DELAY IN MILLISECONDS
PLUS 50% DC M E & TA ILS
700
PLUS 95% DC M E & TA ILS
PLUS 99% DC M E & TA ILS
600
500
400
300
200
100
0
DESTINATION - CARRIER
10
TABLE A - PERCENTAGE OF CABLE/SATELLITE CIRCUITS WITH DELAY (D) IN MILLISECONDS
ARGENTINA
AUSTRALIA
CHINA
FRANCE
GERMANY
HONG KONG
INDIA
ITALY
JAPAN-ITJ
JAPAN-KDD
KOREA-KT
NETHERLANDS
PHILIPPINES
RUSSIA
SINGAPORE
SPAIN
TAIWAN
THAILAND
UK-BT
UK-MERCURY
SATELLITE
D
%
520
26.6
520
8.3
520 100.0
520
25.8
520
23.6
520
11.8
520
54.1
520
4.7
520
29.9
520
14.5
520
22.3
520
15.0
520
4.8
520
25.0
520
55.2
520
25.0
520
13.1
520 100.0
520
31.5
520
12.5
1
2
CABLE ROUTE
3
4
%
D
D
109.7
143.0
%
40.0
91.7
D
138.0
%
33.4
D
62.2
62.2
115.6
140.0
100.0
89.2
89.0
109.0
62.2
122.6
86.7
142.6
62.2
107.8
25.6
21.8
35.4
30.4
95.3
20.5
73.2
33.1
25.0
25.7
75.0
22.4
8.4
13.5
65.9
63.9
155.7
180.0
5.7
7.3
35.4
15.5
68.4
68.4
160.0
42.9
12.7
17.4
148.0
94.7
137.0
66.4
137.7
49.6
12.3
44.6
5.0
23.4
68.4
140.0
146.8
70.4
117.2
22.4
8.3
13.6
62.2
62.2
10.9
23.2
63.9
64.5
16.3
3.6
5
6
%
D
%
71.0
18.2
71.4
16.4
5.0
46.1
71.0
50.0
73.8
143.8
8.3
55.3
80.5
146.2
50.0
4.5
64.5
65.6
3.3
46.4
65.6
66.4
1.1
5.4
66.4
76.0
23.9
8.9
D
%
76.0
13.0