Over time digital techniques have been devel-
oped extensively and hence should replace older
techniques if it were possible to disregard exist-
ing assets. Digital transmission is now cheaper
and exhibits better quality than analog ones. But
several of the existing world-wide telenetworks
are still predominantly analog, represent large
assets and function well. Further expansion and
replacement – analog to digital – imply many
questions, trade-offs and techniques for change
or coexistence.
As the demand for communication channels be-
tween computers and their possibilities increase,
digital transmission and also packet switching
continue to become more advantageous. That is
especially the case for traffic consisting of many
and dynamically changing performance factors.
An example illustrates that:
Some computer co-operating applications are
characterized by burstyness. Machine A sends
one or more messages to machine B. Then for
some time nothing is sent because A has to do
something else, A’s user does some thinking, A
is awaiting answer or acknowledgement from B,
then B may suddenly respond with a large mes-
sage, and so on. Traffic in the communication
channel becomes bursty. Efficiency of such co-
operation often requires short transfer time to
avoid idle waiting of the receiver. This is typical
for answer- and acknowledge-type messages.
The length of various messages typically varies
a great deal. Transfer time for a message
depends on the bandwidth of the channel, i.e. the
bit transfer rate. Large bandwidth gets messages
through fast. This would tend to make relatively
poor use of the channel. It would be idling much
of the time. More bursty traffic means less effi-
cient use of the channel because of more idling
time. A packet switched channel may be used by
a multiplex of packet streams. It may be espe-
cially attractive to mix packet streams with dif-
ferent transfer time requirements. Urgent packets
slip through fast while packet streams of lower
transfer time requirement can wait and make use
of otherwise idle periods.
History of Initial Internet
Development
Development Teamwork
The Arpanet collaboration that eventually led
to the establishment of the Internet was carried
out by a handful of research groups collaborat-
ing intimately for many years. Lawrence “Larry”
Roberts initially led the Arpanet work as director
of ARPA’s Information Processing Techniques
Office – IPTO. Robert “Bob” Kahn later re-
placed him. More than anybody else Kahn was
the person who formulated goals and guided
development of the Internet-technology during
the most active development period. Vinton
“Vint” Cerf later assisted Kahn. Later on, Cerf
promoted and led the further development of the
technology and its applications. Vint Cerf today
appears as “Internet Guru number one”. All
these three persons distinguished themselves as
excellent professionals and were able to moder-
ate and lead the many capable and outspoken
researchers and research groups that took part
in the development.
During the 1970s the following ten groups par-
ticipated in the development of Internet technol-
ogy as one intimately collaborating team:
- Advanced Research Projects Agency – Infor-
mation Processing Techniques Office, Wash-
ington, DC – ARPA - Bolt Beranek and Newman, Cambridge,
Massachusetts – BBN - Stanford Research International, Menlo Park,
California – SRI - University of California, Los Angeles – UCLA
- Information Sciences Institute, Marina del
Rey, California – ISI - Linkabit Corporation, San Diego, California
- Linkabit
- Comsat Corporation, Gaithersburg, Maryland
- Comsat
- Massachusetts Institute of Technology,
Cambridge, Massachusetts – MIT - University College London, England – UCL
- Norwegian Defence Research Establishment,
Kjeller, Norway – NDRE.
Among these groups the company BBN had
many central and decisive roles in the develop-
ment. BBN was responsible for the daily opera-
Line x
Line y
Line y
A to B
B to A
Other traffic
in between
Packets
Dialog
In a packet switched network
each leg may carry a mixture
of traffic streams belonging
to several different “con-
versations”. Each packet is
addressed and has necessary
identification with it