|
The Digital Future Of The
Telephone Network
A Study of Evolving Technology
By Lee Goeller
Originally published by Probe Research Inc. 1979.
Reprinted by permission
Chapter 8
All Those Other Networks
Switching was invented
to reduce the cost of transmission; it is less expensive to give
everybody a pair to one central point than to give everybody a pair
to everybody else. However, over the years, the cost of switching
has increased while the cost of transmission facilities has tended
to go down: automatic switching costs more than a manual
switchboard, common control systems cost more than SXS, stored
program systems cost more than electromechanical, etc. Thus one item
of cost reduction in rendering telephone service is the elimination
of switching. If a large business customer has a lot of traffic
between two of his locations, direct tie lines, bypassing the public
network, relieve the toll hierarchy and, in particular, the toll
switches, of a lot of work. Thus tie lines save the telephone
company money, and, assuming they are priced properly, they can make
a profit, give the business user a chance to cut his costs, and
reduce the capital required for switching in the public network.
When a customer has
three or four locations, he may start building a network, switching
tie lines through his switchboard. Dial tandem networks, implemented
with extra SXS switches added as required on SXS PBXs, got a great
shot in the arm with the coming of Telpak, a bulk tie line offering
designed to make telco-provided facilities more economically
desirable than private microwave. Many corporate networks sprang up,
just for voice traffic. For larger customers, CCSA (Common Control
Switching Arrangement) was developed later and used central office
equipment for tie line switching. The passing of SXS and the high
cost of CCSA have led to new Dimension PBX offerings (ETS, in this
case electronic tandem switching and NOT electronic translator
system as in No. 4 Crossbar) which will keep private voice networks
viable. Indeed, tie line networks may increase.
The so-called
Specialized Common Carriers are largely in the business of providing
analog long-haul tie lines in competition with AT&T. Apparently the
FCC authorized the SCCs to be sure that the expected "data
explosion" would find enough trunks available; unfortunately, most
of the SCCs have gone in with traditional analog microwave radio
systems that seem to be used primarily for voice and, indeed, are
sometimes reported by users to be unsatisfactory for high speed (9.6
kb/s) data. In any event, low cost trunks from SCCs are also
encouraging private networks; these networks are as non-digital as
the AT&T long-haul trunks, and for exactly the same reasons.
What, then, of data?
Surely there is some out there, and it is being handled some way or
other. What is actually happening is that modems and acoustic
couplers are handling data on voice-frequency facilities. An
elaborate hardware business has grown up providing multiplexers and
concentrators to allow one voice frequency facility to handle a
number of data streams simultaneously, and to allow a number of
terminals, many of which may be idle at any given time, to have
access to a somewhat smaller number of data channels. Small
computers built into such equipment give it an incredible degree of
flexibility and capability.
Thus data networks,
using both private voice tie lines and the public switched network
(as well as private switched voice networks), are in wide use and
increasing rapidly. Switched data networks are a little harder to
come by. Analog data trunks, particularly those at higher speeds,
must be "equalized" to prevent transmission distortions (which do
not affect voice communication appreciably) from making the signal
unreadable at the far end. But connecting two arbitrarily selected
data trunks back to back (as in voice switching) degrades the
overall equalization. Thus store and forward techniques, culminating
in "packet switching," have been developed.
In such systems, each
equalized channel is fixed, and the message is transferred from one
machine to another via the channel. The receiving machine then reads
the "header," sees where the message is to go, queues it for the
next hop, and pumps it out as soon as the required channel comes
free. The queuing insures better utilization of the facilities, and
the switching function is reduced to moving information around
inside a computer and regenerating it on the outgoing channel.
If messages to be sent
are very long, other messages may be delayed considerably. Thus
packet switching breaks long messages up into shorter messages of
uniform length and interleaves such packets in time. In a properly
designed packet switching system, the delays normally associated
with store and forward techniques are avoided and the user feels
that he has a direct connection between, say, his terminal and a
distant computer.
The need for data
systems, particularly for smaller organizations that cannot take
advantage of the savings that quantity can produce, has led to
value-added networks. "Resale of service" has been forbidden in
telephone tariffs for many years, along with rules forbidding the
connection of "foreign attachments." But does a time-sharing
computer service, where customers dial up ports on remote data
concentrators serving remote computers, constitute "resale"?
Apparently not any more. Several value-added networks have sprung up
in the last few years, rendering a useful service that is likely to
increase.
The point in all this is
that many, many networks are springing up for electrical
communication all over the United States. All of them represent, to
some degree, a fragmentation of the overall system. This is neither
good nor bad. It just seems to be happening; if services are
required, there is no longer any need to deny them to customers
because of arbitrary utility policy. But fragmentation is a result.
Within the Bell System,
the "Transaction Network" uses the public switched network for
low-density traffic, but uses standard data techniques for traffic
at high densities. Designed for very short holding-time connections
(credit card checks, etc.), the transaction network has traffic that
can be greatly benefited by CCIS. A 10 second call set-up time isn't
too bad for a 5 minute call, but for a 3 second call, it is
preposterous. The transaction network would seem to be an ideal
example of the sort of thing that could be incorporated directly
into the public network if only that network, with its high-speed
CCIS signaling, were end-to-end digital.
Bell's Digital Data
Service (DDS) is still not switched. It constitutes a true digital
network, using T-carrier in its digital form; only when customer
access requires a voice loop to the nearest T terminal do modems
appear. To get between major cities where T-carrier is not normally
used. Data Under Voice (DUV) has been developed. DUV uses the radio
carrier already present for analog long-haul voice trunks. It turns
out that, in modulating the groups and supergroups of voice channels
onto the radio signal, there is a portion of the spectrum that is
too distorted to use for analog communications. It works fine for
digital signals, however, and each radio beam can carry one
equivalent of a T span-line (1.544 megabits per second). On fat
routes there may be several radio beams, and 1.544 mb/s can be
subdivided into a lot of 1.2 or 2.4 kb/s data streams. Thus for a
very small incremental cost, large numbers of point-to-point
low-speed data channels (and even some much higher speed channels)
can be offered around the country. The interesting point here is
that DUV cannot help but add to the profits earned by long-haul
analog microwave carrier. Thus a true digital data system will do
its part in delaying the development of a universal digital network.
It is unlikely, of
course, that a universal digital network would (or even should)
eliminate all other networks. However, the impact of fragmentation
on customer costs and on frequency spectrum available for radio
channels cannot be ignored.
[
Top ] [ Table of Contents ] [
Next Chapter ] |
