10 / 25/95

09:15

©703 620 0913

@002

THE INTRODUCTION OF PACKET SATELLITE COMMUNICATIONS

Robert E. Kahn

Advanced Research Projects Agency

Abstract

The preparations which led to the SATNET
Experiment are discussed in this paper. Various
packet satellite tariff considerations and
architectural issues are presented along with a
summary of future plans for use of the technology.

1. Introduction

This paper reviews the process which led to the
introduction of packet satellite technology in the
1970s. The development of this technology was
undertaken by the Advanced Research Projects
Agency (ARPA) In order to evaluate its utility for
efficient long haul computer communications with a
potentially large number of geographically
distributed users. This effort was undertaken in
conjunction with participating organizations in
the U.K. and Norway, but does not necessarily
reflect their views on this subject.

The''most notable example of this technology is the
Atlantic Paoket Satellite Network, known as
SATNET, which has been in operation on the
Atlantic Intelsat IV satellite since late 1975 and
which currently serves a community of researchers
in the U.S, the U.K. and Norway. Underlying the
SATNET technology is the basic packet switching
technology which was first Introduced during the
late 1960’s. The November 1976 IEEE Proceedings
contains a comprehensive treatment of packet
communications technology and includes a paper on
General Purpose Packet Satellite Networks which
provides a good introduction to the subject [1,2].

SATNET consists of a^^single broadcast channel
shared by multiple earth’stations which use Time
Divieion Multiple Access (TDMA) and emit packets
according to a channel access protocol. These
earth stations may be connected to one or more
aubsoriber networks. Each earth station contains
a programmable satellite processor (a controller
and related electronics) which implements the
satellite channel protocols and interfaces. The
system provides complete connectivity between all
the participating earth stations and allows
dynamic allocation of the satellite channel among
them. Different priority levels may be supported
efficiently on the same channel without
unnecessary preallocation or preemption of

resources. The broadcast property of the channel
enables a transmission from one earth station to
be received by all the others including Itself.
Both conferencing and delivery of multiple address
packets can be achieved efficiently as a result.

The Arpanet was the first example of a packet
switched network which used point-to-point
terrestrial lines (across the U.S.) in a store and
forward system 13,**]. Following the installation
of the first Arpanet nodes, a number of papers
appeared in the literature on the application of
packet switching to multiple access radio channels
[5,6,7,8]. The ARPA-sponsored effort at the
University of Hawaii was the first to demonstrate
burst transmission of packets by radio for
computer access by terminals within line of sight
of the computer center. In this system, called
the ALOHA system [9,10], packets were simply
transmitted when they were ready to send - at
random instants of time. No explicit control of
the radio channel was invoked. Rather, on
occasion, packets would collide in the air,
destroying each other and would be retransmitted
at a later random time. The multiple access
nature of this system resembled a packet satellite
net, except that the terminals were much closer to
and quite unequally apaced from the computer
center which (like a satellite) formed the hub of
the system. The Hawaii researchers extended the
concept of radio packets to satellite
communication directly, and experimentally
verified the concept using test packets over
NASA’s ATS-1 satellite between Hawaii and
NASA-Ames. The technique of operating a Packet
Satellite Net in an uncontrolled fashion became
known as "Pure Aloha".

A significant body of theoretical work on the
analysis of Aloha Systems appeared in the
literature in the early 1970's and various
improvements on the original random transmission
technique were proposed and evaluated [11,12]. In
the Slotted Aloha technique, first Introduced by
Roberts, the time axis at the satellite is divided
into equally spaced Intervals called time slots
which hold a single packet [133- Under the
Slotted Aloha regime, packets can only be
transmitted starting at the beginning of a slot.
For fixed length packets and Poisson traffic
arrivals, the capacity of the slotted system is
twice that of the unslotted system due to the
reduoed number of collisions at light traffic

10 / 25/95

09 :17

©703 620 0913

@003

2

loads. In both the slotted and the unslotted
ease, some form of stability control Is needed
[19,15). For efficient use of a packet satellite
channel, where the ane-way~tfranslt tine is ouch
larger than the-time to send a packet, aooe form
of satellite channel allocation strategy is
appropriate C 1G3 . A priority oriented demand
assignment technique was developed for dynamic
allocation of capacity and is currently in daily
use in SATNET [17}.

Both simulation and analysis were used extensively
and effectively in investigating these and other
channel access schemes. However, this work was
unable to deal effectively on a purely theoretical
basis with any of the practical problems
associated with development of the technology.
Access to an experimental system was essential to
address topics such as fault detection and
isolation, system status monitoring and debugging,
interfacing to terrestrial networks and gateways,
software structure and performance. It was
feasible to carry out a test of the technology on
one of several existing satellites and it appeared
as if existing ground stations could be used in a
packet mode of operation with only minor
modifications to provide external on/off control
of the carrier by a programmable satellite
processor at each earth station. The packet
satellite technology was also seen as a
potentially useful long term adjunct to existing
network technology for long haul applications
involving conferencing, multi-destination
broadcasting and especially to provide
connectivity between a large number of sites (each
with low duty cycle traffic) using a small
fraction of the leased channel bandwidth that a
fully connected network of point-to-point cirouits
would have required.

In the 1973—197-4 time frame, the only viable
choices for such a test (from the U.S. point of
view) were the Intelaat satellites, the NASA
experimental satellites and one of the several
military satellites. The Intelsat ayatem was a
preferred choice for this activity because it
oould be made available most easily and had the
potential for supporting the resulting technology
on a commercial basis upon completion, if it
proved to be economic. The military satellites
were less appropriate choices as there was not yet
a stated requirement for packet satellite service
in the military. At^hfegaae tide, international
interest in packet switching was growing
significantly, and possible requirements for
interconnection of domestic packet networks in the
different countries were identified. In 1973, the
ARPANET had Just been extended to Norway and the
II.K., and experimental use of the ARPANET was
proving to be quite worthwhile for research
purposes.

This is the context in which the subject of an
experimental program on packet satellite
technology was first raised with the British Post
office, with the Communications Satellite
Corporation (Comsat), and subsequently with the

Norwegian Telecommunication Administration (NTA)
-and Norwegian Defense Research Establishment
(NDBE). Jn the following section, the
preparations for the SATNET experiment are
outlined along with the approval process which was
required.

2. Preparing for the SATNET Experiment

In 1974> -the U.K. -Poet Office agreed to support
the SATNET experiment by contributing the U.K.
half of the satellite link and by providing access
to the necessary earth station equipment in
England, a programmable satellite processor was
installed at the Goonhilly earth station and
connected back to a gateway at the ARPANET node on
the premises of University College London (UCL)
with a 96 Kbps communication line. UCL was
prepared to accept the main research
responsibility for the U.K. participation in the
SATNET program, and subsequently did so.

Also in 1979, Comsat agreed to U.S. participation
in such an experimental activity, but only if it
were carried out under the auspices of one of the
several U.5. International Record Carriers (IRCs)
which historically have played the role of
intermediary in bringing international data
services to the end customer. Comsat Is the U.S.
representative to Intelsat. When the SATNET
project was being formulated, Comsat also operated
both the space segment under contract to Intelsat
and the U.S. earth stations for the consortium of
U.S. owners, Intelsat itself now operates the
space segment.

The only generic classes of service which could be
offered by Comsat were those specifically approved
(tariffed) by Intelsat. Clearly, a packet
satellite service was not among them. After a
period of discussion within Intelsat lasting
several months, an Intelsat tariff for a
multi-station service was approved in late 1979.
The SATNET program was initiated in September 1975
with one Intelsat standard A (30 meters) earth
station at Etam, West Virginia and a similar one
at Goonhilly Downs, England. Within Norway, the
Interactions with the NTA were handled entirely by
the Norwegian Defense Research Establishment.

While Norwegian participation in the SATNET
program had started with the first meeting of the
researchers in 1975, their active participation on
the channel began in late 1977 using the Nordic
earth station at Tanum, Sweden. Shortly
thereafter, Comsat Laboratories made preparations
to participate actively with a small Unattended
Earth Terminal (UET) at Clarksburg, Maryland for
system diagnosis and evaluation. The UET differed
from the three standard A earth stations In that
it had only a 10 meter antenna and could only
receive at 16 Kilobits/second while the other
stations could receive at 69 Kilobits/second. All
four stations can transmit at 69 Kiloblts/second,
but the large stations rau3t reduce their
transmission rate to 16 Kilobits/second to talk to
the UET.

10 / 25/95 09:18

©703 620 0913

® 004

3

The technical aspects of the SATNET experiment are
not addressed in this paper. Other companion
papers address both system level and experimental
aspects of the program [7tf, 19.20,21 ] . In the
remainder of this paper, the relevant tariff
considerations will be discussed and two key
architectural Issues will also be considered.

3. Intelsat Tariffs

The new Intelsat tariff which was approved In late
197 M was for a new kind of service known as
Hulti-Destinatlon Half-Duplex (MDHD). Simply
stated, KDHD allows one or more members of
Intelsat to jointly share a common channel on any
of the Intelsat Satellites for a modest MDHD
payment to Intelsat. The normal leased service
offerings from Intelsat to its members are a
point-to-point service between two earth stations
and a broadcast service from one prespecified (but
fixed) earth station to at least two others. The
point-to-point service can be either half-duplex
(one way) or full-duplex (two-way). The broadcast
service utilizes only one channel, as a reverse
path is not included.

The MDHD capability may be viewed as an extension
of the broadcast service to allow more than one
prespecified earth station to transmit. MDHD
allows all participating earth stations to
transmit using their own channel access protocol
to resolve contentions.

To any member country already participating in an
MDHD service, the added cost is nominally zero to
allow additional earth stations to share the MDHD
channel. This assumes that capacity limitations
are not exceeded and that coordination among a
larger number of sites costs the same. However, a
payment must be made to Intelsat by each member
country which chooses to join (share) an existing
MDHD channel, so the total payment received by
Intelsat for MDHD servioe grows linearly with the
number of countries. The reasons for a choice of
tariff in which the cost per country is
independent of the number of participants depends,
in part, upon the political structure of Intelsat.
The subject of PTT tariffs to the end customer,
although not specifically discussed In this paper,
would generally include earth station and
terrestrial charges, as well as space segment
charges.

If we assume Intelsat normally charges a member C
per one-way channel of a certain capacity for a
total of 2C counting both ends, then the same
revenue would be gathered if each of the
participating members in an MDHD channel were
charged 2C/N apiece (assuming N participants).

Tne members, in turn, could base charges to their
customers on these costs plus the added costs of
ground station equipment and terrestrial
interconnection. Thi3 kind of formula in which
the space segment chsrge is independent of the
number of earth stations appears well-suited to
domestic services where all the earth stations are
owned by one authority. However, this formula :

poses several problems when applied to the
'international situation, .where the earth stations
are separately owned and operated.

First, the rate -base for -each participating
country would fluctuate as a function of the
number of participating countries, making
financial management and planning awkward and
unpredictable at best. Second, and more
important, -the voting rights of each member
country in Intelsat are a function of its total
payments to Intelsat. Primarily, for that reason,
the Intelsat MDHD tariff was fixed to be a
constant C per channel per country.

The Intelsat broadcast tariff illustrated in Fig.
Ha) shows one transmitter which is charged C and
four receivers each of which is charged C/2 for a
total of 3C. 5ince at least two receivers must be
present for a broadcast service, the minimum
charge is 2C (which is identical to the half
duplex point-to-point tariff between two sites).
The revenue produced by the broadcast tariff
increases linearly with the addition of more
ground stations at an increment of C/2 per added
receiving station.

The MDHD tariff illustrated in Fig. 1(b) shows
each participating country being charged C for the
right to receive and transmit on the same channel.
The net payment to Intelsat, 5C, is almost double
the charge for the simple broadcast case.

However, the value received for this added cost is
full N-way connectivity since any of the earth
stations can transmit to the others at any time
according to the chosen channel access protocols.
The MDHD tariff is also considerably cheaper than
that for N distinct broadcast channels to
implement N-way connectivity (NC vs. [NC -t-
N(N-1)C/2]). Along with the initial higher cost
of N broadcast channels would come N times the
capacity, however, regardless of whether it can be
used effectively or not.

With these existing tariffs, the cost per country
normalized by total number of channels of network
capacity available to the N earth stations is C
for the MDHD case and [C + (N-1)C/2]/N = (N+DC/2
for the case of N broadcast channels. If existing
MDHD tariffs are extended to channels with a
higher bandwidth using a "linear extrapolation" of
the current tariff, the charges for obtaining the
added capacity with multiple lover capacity
broadcast channels will be half as much as the
single MDHD channel as the number of participating
earth stations becomes large. Since this ratio
reflects only the current tariff structure, the
ratio could be changed (e.g., become closer to
unity) with a non-linear tariff revision
applicable to higher bandwidth channels.

From an architectural point of view, the use of
multiple broadcast channels has both positive and
negative features which are identified in section
6. However, In most applications, it is doubtful
whether initial network-wide traffic will be large

10 / 25/95 09:20 ©703 620 0913

® 005

4

enough to Justify commencing service with more
than a single MDHD channel.

4. COMSAT and IRC Filings in the 0.5.

The U.S., U.K. and Norway participation in SATNET
has bsen on an experimental basis and a service
has not- yet been offered to customers in any of
these countries. In the U.S. , Comsat filed a
tariff with the FCC in 1975 to offer an
experimental packet satellite capability to ARFA
and its designated contractors via one of the
IRCs. The service Comsat offered was based on the
HDHD service obtained from Intelsat, and was
augmented as required with the programmable
satellite processor at the earth station. In its
filing, Comsat also referred to its augmented
service as MDHD.

Comsat bought or leased all the necessary ground
station equipment to provide the experimental
service as for a normal oommercial service
offering. In a competitive selection. Western
Union International (WUI) was chosen by the U.S.
government from among the IRCs to provide the MDHD
service to the ARPA program. WUI, in turn, filed
a tariff with the FCC to offer MDHD service as
obtained from Comsat, which they augmented with a
terrestrial circuit before supplying it to tbe
government.

The statement of work which accompanied the
government's request for proposals was unusual in
that it did not cite any specific destination or
customer location abroad. Rather, it simply asked
for an MDHD channel from the U.S. to an
unspecified point in the U.K. and stated that all
of the U.K. costs were to be assumed by the
British Post Offioa. A point of contact in the
Post Office was identified. The request also
stated that additional unspecified destinations
might have to be connected subsequently, as Norway
eventually was.

To validate the initial delivery of the service
and to verify reatoral of service in the event of
outage, only a loopback test from the U.S.
customer site (which was specified to be the
Seismic Data Analysis Center in Alexandria,
Virginia} to the satellite and back was required.
The payment to VUI was not dependent on the
participation (or performance) of amy other
country (or its equipment!. . However, the WUI and
COMSAT tariffs both included a small charge
proportional to the number of participating sites
for coordination.

A diagram of the MDHD payment flaw during the
experiment is shown in Fig. 2.

5- SATNET Experiment

The SATNET Experiment was conducted nominally
during the period from September 1975 through
September 1978 and involved researchers from each
of the three participating oountrles. The basic
physical architecture of SATNET was dictated by

many programmatic considerations (e.g. use of
existing ground stations and satellites) so the
actual hardware configuration merely reflects what
was available for use in the experiment. However,
the logical architecture of the system has been a
subject for research and bas constantly evolved
during the oourse of the program. Neither the
software architecture nor the system protocols
were preaorlbed in advance and the non-hardware
parts of the system interfaces were also allowed
to evolve, which they did. Each was a major
subject for investigation and exploratory
development during the course of the project. The
resulting logical architecture will be very useful
in designing a more advanced follow-on system. In
addition, an effort was undertaken to develop and
demonstrate a high performance digital burst modem
and error control unit for possible operational
use with SATNET after the experiment.

A major decision in the program was to separate
the SATNET development and testing from the
closely related internetting research activities
which were Just getting underway. It was decided
to pursue the internetting research using a
separate minicomputer gateway in each country
simultaneously connected as a Host on SATNET and
as a Host on the Arpanet [22,23]. This
arrangement left enough flexibility to pursue
gateway related research without requiring
software changes (in reel-time) to SATNET or
Arpanet. The gateway software could have been
Incorporated within the physical confines of
either SATNET or Arpanet, or split between them.
However, keeping it separate for the purposes of
the experimental program provided maximum
flexibility to the internetting researchers, many
of whom were also working on SATNET, Arpanet or
other ongoing network related programs without
unnecessarily distracting those SATNET researchers
who did not need to be deeply involved in the
internetting work at that time.

Technical direction of the program beginning in
September 1975 was the responsibility of Linkabit
Corporation, San Diego, California who prepared a
comprehensive test plan to guide the conduct cf
the experimental program. Hajor participants were
Comsat, Bolt Beranek and Newman, University of
California at Los Angeles (UCLA) and the Defense
Communications Agency in the U.S., University
College London and the Post Office in the U.K.,
and NDRE and NTA in Norway.

Coordinating a program Involving participants from
multiple countries was an important challenge that
was met at several different levels. Quarterly
review meetings were held (rotated among the
different locations) and attended by all the
participants. Technical progress was reviewed at
these meetings, technical issues were discussed
and resolved and plans for each succeeding quarter
were refined. Research issues end results were
documented and circulated in a series of informal
working group notes. The ARPANET played a
particularly important role in executing the
effort as well as in coordinating it. It provided

10/25/95

09:21

©703 620 0913

® 006

5

the means by which the satellite processors were
down-line loaded and debugged, and the Beans by
which SATNET itself was controlled and monitored
as it was being developed. The message passing
capability of the hosts on the ARPANET were used
to Weep all participants informed of technical
progress, system status, often by direct reporting
from the programmable satellite processors in
SATNET, and to resolve questions and coordinate
experiments on a day-by-day basis. Without such a
capability, it is doubtful that the overall
experimental program could have been carried out
successfully.

The main results of the experiment are being
documented by Linkabit Corporation (with inputs
from all the participants) in a final report to be
available shortly. A summary of the findings show
that the SATNET experiment demonstrated the
feasibility of the packet satellite technology,
illuminated many of the most important technical
and non-technlcal issues and provided a system
that can support advanced computer communication
research applications. Although the subject of
packet voice has not been emphasized in this
article, it played an integral role In the SATNET
design. SATNET is the only operating long haul
packet switched network in the world that has been
designed to handle both packet switched voice and
data.

6 . Architectural Issues

Two architectural issues arose during the course
of this project which are appropriate to single
out for mention. The first issue is selecting the
functionality that ought to reside in the
processor which is colocated with the rest of ths
earth station equipment and the functionality that
ought to reside at the terrestrial interface (to
the earth station) which might be located some
distance from it. The second major issue concerns
the means of increasing the overall traffic
handling capacity of the system. Each of these
issues are briefly mentioned below.

a. Functionality of the Earth Station and Its
Terrestrial Interface

Although not all the functions implemented in
SATNET need to resiftsugl^the earth station, a
minimum set of functions must be located there to
control timing and acoess to (and transmission on)
the satellite channel. Farts or the functionality
might be moved to a terrestrial location distant
from, but connected to, the earth station by a
communication line. One attempt at the definition
of the functionality is given in [24]. In
particular, certain aspeots of the functionality
which deal with multiplexing traffic from many
users into a composite stream to the earth station
could probably be relocated without penalty In
performance provided delay or unreliability is not
added outside the earth station. Accounting and
other administrative functions could also be
remoted from the earth station without penalty.

b. Expansion of Network Capacity

Although a single 64 Kilobits/second channel is
utilized In the SATNET experiment, this capacity
clearly would -be insufficient for many
applications. The capacity of a SATNET system
could be expanded la several ways. First, it
could be simply scaled up in data rate. The
ability of s packet switch to handle,
multi-aegablt/second data has been demonstrated
[25]. This would suffice for an expansion of one
or two orders of magnitude. A transponder car.
typically handle upwards of 60 Megabits/second,
however, and the newer satellite systems are
expected to support higher data rates still.
Multi-processor systems seem capable of supporting
these higher data rates on a single shared
satellite channel without either increased delay
in buffering or processing. However, the number
of processors must grow linearly with capacity and
special attention must be paid to communication
between processors and with external devices.

A second alternative, which becomes attractive
when the overall network traffic is high enough,
is to incorporate dedicated uplinks using
Frequency Division Multiple Access (FDMA). In
this scheme, which is illustrated in Fig. 3, a
separate processor at each earth station would be
dedicated to handling traffic on each broadcast
downlink and would pass along to a concentrator
only those packets destined for its earth station.
The capacity of the concentrator would then be
sized to the throughput intended for that site
which presumably would be much less than the total
network traffic. In this scheme, a modification
would be required at each existing ground station
for each new addition to the net, which is a major
disadvantage. However, it is highly modular and
should be easy to upgrade in an operational
system.

The use of multiple FDMA broadcast channels, one
per site, reduoes the earth station processing
requirements but it also does not provide the
flexibility that comes from the dynamic assignment
of capacity in a TDMA system. A third alternative
is a hybrid of cases one and two sbove in which
some of the uplinks may be MDHD channels (using
TDMA) while the rest may be broadcast channels
each from a single source.

7. Future Plans

SATNET currently serves as the backbone for a
number of innovative research applications and has
become the primary packet transportation vehicle
between the U.S. and Europe for computer
communications and command and control research.
Since May 1979 , ARPANET access from the U.K. has
run almost exclusively via SATNET on a provisional
basis. It is planned to continue the use of
SATNET as the primary link between ARPA and its
research partners in Europe. The ARPANET link to
London (via Norway) is scheduled to be taken down
during the last quarter of 1979 after which the
only available ARPANET access from the U.K. will

10/25/95 09:23

©703 620 0913

@007

6

be via SATNET. NDRE Will utilize SATNET for
research purposes; the only planned use of the
remaining point-to-point ARPANET link from the
U.S. to Norway will be foi^retrieval of seismic
data, which was the original function of that line
prior to its incorporation in the ARPANET in 1973-

Within the U.S., ARPA plana to use the SATNET
technology as the basis for a domestic packet
satellite channel operating at 3 Mbps with 5 meter
antennas initially at Lincoln Laboratory, in
Lexington, Massachusetts and USC/I5I in Marina Del
Rey, California. Additional sites in the San
Fransisco area and Washington, D.C. area will
also be added. A major purpose for the domestic
channel is to explore the use- of the SATNET
technology to support multi-user integrated packet
voice and data communication. Only with the
increased bandwidth will a test of this concept be
possible using multiple speech and data sources
Including a mix or 2.A Kilobits/second to 6A
Kilobits/second speech transmission with graphics,
facsimile and normal computer to computer traffic.
The Defense Communications Agency also plans to
utilize this technology along with ARPA for
advanced research and development on DoD
integrated data/voice networks of the future.

On the international scene, packet satellite
technology may be useful for a wide variety of
potential applications. One such possibility
which is being offered as a service by the PTTs
and the US Postal Service is Xntelpost. This is
an innovative new facsimile service which is being
tested among several countries. If the current
plans evolve, individual point-to-point links
might be required between each participating
oountry. A packet satellite system could support
the initial Intelpost traffic with only one shared
channel, with considerably less total satellite
bandwidth than multiple point-to-point links would
require and without noticeable degradation of
performance.

Acknowledgments

Tnis effort would not have been possible without
the cooperation and support of the British Post
Office and the Norwegian Telecommunications
Administration; both organizations played a very
central role in the program. Bolt Beranek &

Neuman (BEN), COMSAT, and Linkabit Corporation
played significant hglsfcj-n developing the packet
satellite technology.'" COMSAT spearheaded the
approval process. UCL, NDRE, UCLA, and COMSAT,
with the assistance of BEN, carried out the
measurement, testing and applications developments
despite the large geographic distances from SATNET
and each other which might otherwise have been a
deterrent. The success of the program was due in
no small measure to the technical direction
provided by Linkabit Corporation.

References

[1] Special Issue on Packet Communications, IEEE
Proceedings, Nov. 1978

[2] I.M. Jacobs et al, "General Purpose Packet
Satellite Networks," IEEE Proc., pp. 14AS -
1A67, Nov. 1978

[3] L.G. Roberts and B. G. Wessler, "Computer
Network Development to Achieve Resource
Sharing," AFIPS Conf. Proc., SJCC, pp.

5^3-5*19, 1970

[A] R.E. Kahn, "Resource Sharing Computer

Communication Networks," IEEE Proc., pp. 1397
- 1A07, Nov. 1972

[53 N. Abramson, "The Aloha System - Another
Alternative for Computer Communications,"

AFIPS Conf. Proc., FJCC, pp. 695 - 702, 1570

[ 6 ] L.G. Roberts, "Extensions of Packet
Communication Technology to a Hand Held
Personal Terminal," AFIPS Conf. proc., SJCC,
pp. 295 - 298, 1972

[ 7 ] H. E. Kahn, "The Organization of Computer
Resources into a Packet Radio Network," IEEz
trans. on Comm., Vol. C0H-25, PP- 169 “ 178,
Jan. 1977 (also in AFIPS Conf. Proc., NCC,
pp. 177-185, 1975)

[S] L. Kleinrock and F. Tobagi, "Random Access
Techniques for Data Transmission over Packet
Switched Radio Channels, AFIPS Conf. Proc.,
NCC, pp. 187-201, 1975

[93 R. Binder et al, Aloha Packet Broadcasting -
A Retrospect," AFIPS Conf. Proc., NCC, pp-
203-215, 1975

[10] N. Abramson and F. Kuo, Editors, Computer
Communication Networks, Prentice Hall,
Englewood Cliffs, N.J., 1973 (see chap, on
the Aloha System)

[11] N. Abramson, "Packet Switching with
Satellites," AFIP5 Conf. Proc., NCC, pp.
695-702, 1973

[12] L. Kleinrock and S. S. Lam, "Packet Switching
in a Slotted Satellite Channel," AFIPS Conf.
Proc., NCC, pp. 703-710, 1973

[ 13 ] L. G. Roberts, "Aloha Packet System with and
without Slots and Capture," ACM SIGC0KM,
Computer Communication Review, Vol 5, No. 2,
April 1975

[IK] L. Kleinrock and S. S. Lam, "Packet Switching
in' a Multiaccess Broadcast Channel:
Performance Evaluation," IEEE Trans, on Comm,
Vol. COM-23, PP- A10-A23, 1975

[15] 5. S. Lara and L. Kleinrock, "Packet Switching
in a Multiaccess Broadcast Channel: Dynamic

10/25/95

09 : 24

©703 620 0913

@008

7

Control Procedures," IEEE. Trans, on Comm.,
Vol. COM-23i Sept, 1975

[16] I,. G. Roberts, "Dynastic Allocation of
Satellite Capacity through Packet
Reservation," APIPS Conf. Proc., NCC, pp.
711-716, 1973

[17] I.H. Jacobs et al, "C-PODA - A Demand
Assignment Protocol for SATNET," Fifth Data
Communications Symposium, Snowbird, Utah,

1977

[18] I.M. Jacobs et al, “Packet Satellite Network
Design Issues," Proc. NTC, Nov. 1979, in this
Proceedings

[19] P. T. Kirstein et al, "SATNET Applications
Activities," Proc. NTC, Nov. 1979, in this
Proceedings

[20] D. A. McNeill et al, "SATNET Monitoring and
Control," Proo. NTC, Nov. 1979, in this
Proceedings

[21] W.W. Chu et al, "Experimental Results on the
Packet Satellite Network," Proc. NTC, Nov.
1979, in this Proceedings

[22] V. G. Cerf and R- E. Kahn, "A Protocol for
Packet Network Intercommunication," IEEE
Trans, on Comm., Vol. COM-22, pp. 637-648,

May 1974

[23] V.G. Cerf and P.T. Kirstein, "Issues in
Packet Network InterconnectionIEEE Proc.,
Vol 66, No. 11, pp. 1386-1408, Nov. 1978

[24] E.V. Hoversten and H. L. Van Trees,
"International Eroadcast Packet Satellite
Services," ICCC-78 Conf. Proc., Kyoto, Japan,
Sept. 1976

[25] S.M. Ornstein et al, "Pluribus, A Reliable
Multiprocessor," AFIPS Conf. Proc., NCC, pp.
551-560, 1975

10/25/95 09:24

©703 620 0913

@009

8

2 2

Fig. 1. (a) Broadcast Tariff, (b) MDHD Tariff

NORWAY

Fig. 2. MDHD Payment P’lov

BROADCAST DOWNLINKS

BROADCAST

UPLINK

Mil
□ □□□
fill

PACKET

PROCESSORS

i

□

I

CONCENTRATOR 6 INTERFACE

TT

TO TERRESTRIAL NETWORK

Fig. 3. Earth Station Configuration for Multiple Broadcast
Channels in a High Capacity System