|
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 2
The Interaction Between Transmission And Switching
Introduction
To predict where we are going, we have to
know where we are coming from. And when it comes to the telephone
network in the United States, we are coming from a long way back.
Telephony is older than electric power systems, automobiles,
airplanes and many other things we take for granted. It generated
directly talking movies, radio voice-transmissions, TV networks and
other wonders as side effects during a continuing program of
research and development. But all the time, the main thrust in
telephony was the conquering of distance. That we can now talk to
almost any point in the world is a tribute to the success of the
industry, and, in very large measure, Bell Telephone Laboratories.
DDD Development
Long distance calling went through three
stages: operator handled calling, operator dialed calling, and user
dialed calling or direct distance dialing (DDD). DDD began in the
early 1950s, and it is hard, now, to remember a time when it wasn't
standard. Yet the problems that had to be overcome to make it
possible were so vast that it is hard to appreciate how they could
be solved at all.
Consider. Each telephone has to have a unique
number to permit anybody, anywhere, to call it without ambiguity.
Trunks have to be available from any local central office to any
other, and they have to be designed to connect together, back to
back, in an almost infinite number of ways and to work regardless of
the combination. Further, the trunks have to be designed to
standards so that a connection via one route will not be markedly
different from a connection between the same two users via a
different route.
Then, too, trunks have to be provided in
large enough quantities to insure good service, but not in
quantities so large as to be uneconomical. In the transition from
operator handled calls to operator dialed calls, queuing had to be
replaced with an elaborate scheme of automatic alternate routing; by
having machines choose over several possible routes, queuing was
eliminated with the same traffic going via essentially the same
number of circuits. When DDD took over from operator dialing,
automatic alternate routing became even more important since the
caller had to be completely insulated from the process.
Finally, maintenance posed real problems.
When operators placed calls, either manually or by dialing, they
could check on the quality of the circuit and mark unsatisfactory
trunks for the attention of the maintenance force. With DDD, other
means had to be found. Even with highly reliable equipment, the
possibility of one or more troubles in a given time interval when
several hundred thousand trunks are involved is significant, to say
the least. Bad trunks have to be found quickly, marked busy to
prevent repeated seizures by automatic equipment, and then repaired
and returned to service. It is harder to do this than it looks, as
MCI and SPCC have discovered.
The Toll Hierarchy
In any event, trunks are voice channels that
interconnect switches. Switches are arranged in a five-level
hierarchy where each level gathers up traffic from the next lower
level to provide suitable interconnections within a given
geographical area and also to concentrate traffic for access to the
next higher level. At the bottom of the pyramid, there are some
19,000 local or Class 5 central office switches serving customers
directly. These home on an appreciably smaller number of Class 4
offices (about 1600) where, among other things, operators are
concentrated. There are about 300 offices in the Classes 3, 2 and 1,
and they are known, collectively, as Control Switching Points, or
CSPs. At the very top of the hierarchy, there are 10 Class 1 offices
in the continental United States, with two more in Canada.
Only the Class 1 offices have trunks directly
connecting each to the others. However, at lower levels in the
hierarchy, there are trunk groups provided as required by traffic.
These groups are arranged to "skim the cream" off the traffic
between any two points and, when all trunks in any one group are
busy, overflow traffic is bucked to the next higher switch in the
hierarchy. To prevent ring-around-a-rosie routings, traffic always
tries to go direct if possible, and then, if blocked, to higher
level switches. Routing in the originating area goes up the
hierarchy, and in the terminating area, down the hierarchy.
Since traffic that overflows a direct trunk
group shares an overflow trunk group with overflow traffic from many
other groups, high utilization of all trunks is assured for economy
while use of many routes minimizes the probability of a call being
blocked. As automatic alternate routing is arranged, few calls need
more than three or four trunks in tandem between one Class 5 office
and another.
In passing, it should be noted that a
hierarchal network isn't the only possible configuration. It is
simple, effective, and can, with relatively autonomous and simple
switch controls, route calls satisfactorily and without shuttling
back and forth between two or three nodes in a futile
ring-a-round-a-rosie. Within a hierarchy, a switch can receive a
dialed number via an incoming trunk, find an outgoing trunk headed
toward the destination, repeat the called number down that trunk
(perhaps with modifications such as dropping the area code if the
call has progressed to the called area), and connect the incoming
trunk to the outgoing trunk. It can then sit back and relax, knowing
that the next switch will continue the good work and will not return
the call via some other trunk group. With the coming of CCIS (about
which more later in Section VI), other structures may be much more
desirable.
The neat hierarchal diagram shown in
Figure 1 tends to give a false impression of network operations,
however. Things are much more complicated. Some switches may be
combined Class 5 and Class 4, while others may combine several other
classes in one machine. Operator positions are also inserted into
the system in various ways as shown in Figure 2. Access
ranges from a direct termination from a Class 5 switch to a manual
(3CL) switchboard to bridged access via a Traffic Service Position
System (TSPS) to switched access via the Class 4 switch itself
(Traffic Operator Position System, or TOPS, with Northern Telecom’s
SP-1 toll offices).
Figure 1. Switching Hierarchy in the Toll
Network.
Figure 2. Toll Operator Access
In metropolitan areas, "tandem" switches are
used in an unmarked spot between Class 5 and Class 4 offices to
handle local message unit traffic, and also to gather up toll
traffic from several local offices for presentation to the toll
network via one large trunk group rather than several small groups.
Tandem switches have an interesting history.
In the 1920s and 30s, when the Panel switching system was going into
many major metropolitan areas, there might be forty of fifty
thousand telephones served from one CO building. Since a microscopic
percentage of traffic originating in one group of 10,000 lines would
complete to telephones in that same group, Panel switches, following
manual switchboard practice, were split right down the middle: the
originating switches and the terminating switches were completely
separate. The originating switches had a register to record the
called number and a translator that would convert the dialed number
into the signals required to select the route to the terminating
office. A sender would then send digits forward.
With this kind of flexibility (now common in
computer controlled switches but, at the time, quite spectacular
with electromechanical devices), it was possible to rearrange
routing as desired without troubling the calling customers. This
flexibility, too, enabled a number of originating offices to share
one or more stages of switching. "Office Selector Tandem" came about
in this way. The first stage of switching, the "district selector,"
would choose a path to the second stage, the "office selector." The
office selectors would be shared by a number of originating offices,
allowing for great economy in reaching a number of terminating
switches: only two trunk groups, one outgoing and one incoming, to
each office were required, even though there might be several dozen
originating offices sharing office selectors.
Shortly, "Panel Sender Tandem" came along.
Here another register-sender mechanism handled incoming calls,
permitting additional flexibility at the tandem point. Finally,
after crossbar switching took over, Crossbar Tandem was devised.
Crossbar Tandem was ultimately developed into a powerful machine
with much greater routing and signaling capability. But its main
purpose was still to handle intra-local calls in metropolitan areas,
and to provide a single access point to the toll network for many
Class 5 offices.
Just as panel selectors at a common point can
be shared among a number of central offices, so can step by step (SXS)
selectors. Indeed, this is precisely the way dial tandem tie-line
networks are constructed for large business customers with high
inter-location calling rates, and the way switching is handled
outside of the major metropolitan areas. SXS has been around longer
than any other type of automatic switching, and a huge amount of it
is still in service. It has features such as "digit absorbing" which
permit it to do a limited form of automatic alternate routing
without any form of common control and various other features that
make it far more useful than many would suspect. Some SXS systems
have, like the Panel system, been fitted with a
register-translator-sender arrangement to give them increased
flexibility (Director Step), and if it weren't for the space and
maintenance problems involved, such systems could give a modern ESS
a run for its money.
In addition to tandem switches lurking
between Class 4 and Class 5 offices, PBXs sit below Class 5, forming
what some people think of as Class 6 switching systems. Some PBXs
are quite large, and may even be larger than the Class 5 offices on
which they home. And even small PBXs act as concentrators, loading
traffic onto the relatively small number of trunks that connect them
with the CO; a CO trunk from a PBX will, in general, carry five
times as much traffic as a residential or small business line.
Usually, very little PBX traffic originates or terminates in the CO
on which the system homes; it is one of the major mysteries of the
telephone business why Class 5 offices must act like a filter to
impede the flow of calls between PBXs and the rest of the world. It
would be much more logical to terminate PBXs directly on tandem or
toll switches.
In New York this is, as a matter of fact,
often done. Now that direct inward dialing (DID) can be obtained to
a PBX, the old Panel switching systems don't have enough smarts to
feed digits to the PBX to select the extension. Thus Crossbar
Tandems route DID traffic directly into PBXs, bypassing the Class 5
terminating office.
In any event, the public network is seen to
be somewhat more complicated than the hierarchical diagrams would
indicate, but because of the way everything has evolved, it all
works together pretty well. Of the latest additions to the network,
Nos. 1, 2 and 3 ESS, fit right in. Nos. 1, 2 and 3 ESS are, in many
ways, similar to No. 5 Crossbar, the last and most powerful of the
electromechanical Class 5 switching systems. They use reed switch
matrices to connect lines to trunks and, as a result, provide a
"metallic" path that will transmit DC currents, 100-volt AC ringing
signals, and test signals in addition to the small currents that
correspond to voice signals. No. 4 ESS is something quite different,
and we wi11 consider it in some detail later.
(On first reading, the
material between here and the Summary at the end of this section,
which is highly technical, may be omitted.)
E & M Trunks
Now that we can visualize the public network
as a structure of trunks interconnecting switches arranged in a
hierarchal order, we can progress beyond the topological ideas of
links and nodes to the nature of the trunks themselves. Originally,
trunks were simply pairs of wires terminating in jacks and "drops"
at each end. To place a trunk call, the operator would plug into a
jack and crank her magneto generator to send a spurt of ringing down
the pair to the distant switchboard. The ringing would activate the
"drop," a sort of target arranged to unlatch and pivot forward to
give a visual indication at the far end even after the ringing had
gone away. (Magnetos and drops are still used in some military
switches, even in this age of Tri-tac).
The terminating operator would plug in,
resetting and disconnecting the drop, obtain the called number
verbally, and complete the call. With a drop on both ends, such
simple circuits could be used to set up calls in both directions. To
take a call down, another spurt of ringing to "ring off" would be
sent to the "clearing out drop" associated with the cord that
connected the trunk to the line. Simple, effective, and
understandable.
There were, however, problems. Cranking a
magneto was a bother, and sometimes operators would forget to ring
off. Then too, the technology was advancing. Lamps gave much better
indications and speeded call handling.
Thus signaling schemes were developed whereby
the act of putting out lamps cued the operator through the call: a
lamp associated with a trunk would light to signify an incoming
call. The operator would plug in to get the called number, and the
light would go out. The operator would then plug into the called
party and start automatic ringing which would show up as a flashing
light associated with the cords connecting the trunk to the called
party. Upon answer, the cord lamp would go out. If either party
flashed the switch-hook, the cord-lamp would flash at a different
rate to recall the operator. Upon hang up, the cord lamp would light
steadily to request the operator to pull down the cords. Releasing
the line and trunk would put out the cord lamps, and all components
of the connection would be ready for use in another call.
This kind of operation required fairly
complex equipment at the switchboards to make sure the lamps worked
properly. This equipment was concentrated in the trunk circuits
which terminated each trunk, and the cord circuits that were
associated with the cords operators used to make line-trunk or
trunk-trunk connections. The trunks themselves remained simple pairs
of wires, at least for shorter distances.
As distances got longer, following improved
transmission produced by the loading coils invented by Pupin in 1900
and vacuum tubes invented shortly thereafter by DeForest (neither of
whom were employed by the telephone industry), signaling between
switchboards had to be improved. Intricate arrangements of relays,
following the well-known telegraph art, were devised to tell the far
end when the near end was "on hook" of "off hook." This circuitry
had to be matched to the trunk wires, so standard interfaces to the
switchboard trunk circuits had to be constructed. The most common of
these in long distance work was the one using E and M leads. The E
lead told the switchboard what the party on the far end of the line
was doing: was he on hook (E lead open) or off hook (E lead
grounded)? The M lead transmitted local "supervision" to the distant
end. If the local caller was off hook (as when the operator seized
the trunk for his use), the office battery would be connected to the
M lead via a protective lamp. When the party went on hook, this
battery was replaced with a ground.
A certain dissimilarity between the E and M
leads will be noted. E is open or ground while M is ground or
battery, corresponding to on-hook and off-hook respectively. The
general idea was to hit the line with battery to operate a relay at
the far end. Feeding battery or ground directly to the line via the
M lead charged and discharged the line's capacity quickly, and
permitted dial pulses as well as on-hook/off-hook supervision to be
transmitted as SXS automatic switching came into general use. The
only point in going into all this is that today, many generations of
equipment later, E&M lead supervision is still used on the great
majority of toll trunks.*
[*Footnote: In
passing, it should be noted that some people believe
E stands for Ear,
listening to the condition at the distant end of
the
circuit, and M stands for
Mouth, sending out this end's condition.
Although a convenient mnemonic,
history indicates that early
trunk circuits simply had their leads designated
alphabetically,
and E and M just happened to come out that way.]
In any event, with the E&M lead interface,
isolation developed. The signaling sets designed to pass information
over the line became increasingly intricate, and the trunk circuits,
working with the switchboard lamping, followed suit. Both arts
developed separately, sharing only the common interface. As time
went on, carrier systems using electronic techniques put many
conversations on much smaller numbers of wire pairs, coaxial cables,
or radio beams. Because sending battery and ground over such
channels was impossible, supervisory signals were converted to tones
at one end, and back into DC signals at the other.
Early carrier systems included such tone
signaling schemes as part of their design, but as the network grew,
it became common to use broad-band carrier systems such as L (on
coax) or TD-2 (on radio) to go between major centers, and to use
short-haul carrier (N carrier and 0 carrier, typically) to continue
the circuit at each end to specific switching systems. Thus one
trunk, going from a switch at point A to another at point Z, would
be made up of three (or more) carrier systems wired back to back.
Built-in signaling systems, with signaling as well as voice patching
required at each interface, proved uneconomical, and separate
tone-signaling systems were developed to require only one signaling
system per trunk, regardless of its composition. These SF (for
single frequency) signaling systems became quite sophisticated,
since they had to operate within the voice-frequency band and not be
fooled by speech. But they simplified trunk administration and
patching and did their job very well until the Blue-boxers began to
perfect their particular version of electronic fraud.
The point, however, is that we now have three
systems in tandem at each end of a trunk: the switching system with
its trunk circuit, the signaling system, and the trunk facility
itself. Further, when one is dealing with thousands of circuits at
each switch, and making changes daily on trunk routing and group
size, one seldom wires directly from the trunk circuit to the
signaling circuit, or from the signaling circuit to the carrier
terminal. All such equipment is mounted compactly, each type in its
own frame, and cabled out, on a "mass production" basis, to a
cross-connect frame. At cross-connect frames, "jumpers" are run from
the switch's trunk cables to the signaling set cables and, after
picking up the signaling sets, from the cables at the other side of
signaling set to the cables leading to the carrier terminals. There
is another cross-connect frame between the switching matrix and its
trunk circuits, so, from the "outgoing switch" to the transmission
facility, it is not uncommon to have five or six separate entities,
interconnected by office cable and jumpers.
Loop Supervision
E&M signaling is two-way. That is, a trunk
using E&M signaling can be seized from either end to complete a
call. Keeping automatic switching systems from fighting over two-way
circuits when seized simultaneously from both ends leads to
extensive use of one-way circuits, particularly in tandem networks
in metropolitan areas, and elsewhere when economics permit. Where
many offices are within a few miles of each other, as in New York or
Chicago, trunk facilities still need be nothing more than a single
pair of wires, and separate signaling equipment can be eliminated.
In most such circumstances, "loop"
supervision is used. Here the calling switch, like a telephone user
taking his phone off hook, closes a DC current path via the
conductors to the distant office. A relay at that office operates,
and the distant system is made ready to receive the called number.
The number may be sent forward via dial pulses, similar to those
generated by the familiar telephone dial but actually transmitted
from a sender, or via any one of several other forms of signaling.
In SXS areas, the user may actually be controlling the dial pulses
directly (through pulse-regenerating relays) on the link from a
Class 5 to a Class 4 office, omitting the need for a sender.
A trunk using loop supervision has two ends:
an originating and a terminating end. The originating end closes the
loop, and flaps it to send dial pulses. The terminating end contains
a relay that operates upon seizure, and releases momentarily to make
the trains of dial pulses clean and sharp. Battery for the loop is
applied at the terminating office.
We need something more, however; we have to
know when the called party answers and hangs up so that charging,
controlled by the originating Class 5 or Class 4 switch, can be
accurate and effective. To do this, the battery is "reversed" in the
incoming trunk circuit when the called party answers. That is, a
relay actually transposes the leads to battery and ground, and
current flows in the opposite direction over the conductors. At the
originating end, a "polar" relay (one that operates when current
flows through it one way but does not operate when current flows the
other way) detects the reversal to start charging for the call and,
at the end of the call, releases to stop call timing. The polar
relay will only work when the loop is closed, so the terminating end
cannot call the originating end by reversing the battery when the
loop is open.*
[*Footnote: A
form of signaling called "high-low" will
do this, however, but it is seldom used today.]
The modern electronically oriented reader,
aware of devices invented since the relay, may be wondering if I am
stuck in the past with the above references. I am not. I am making a
point. Right up through the development of No. 1 and No. 2 ESS, in
the 1950s and 1960s, loop supervision was standard and basic. Most
short-haul trunks (which is to say most trunks) used it and,
following the approach of No. 5 Crossbar, No. 1 ESS was optimized
for loop trunks. The enormous increase in T-carrier which will, in
the next few years, vastly diminish trunk transmission facilities
consisting of wire pairs in cable, hadn't been anticipated. Thus an
ESS pretends T and other modern carrier systems are cables, and
these transmission systems, or separate SF signaling units designed
to simulate loop operation, maintain the charade from their side of
the interface.
Start-Pulsing Signals
But we have not yet reached the end. All
common control offices tend to be slow. They have to detect an
origination, whether from a line or an incoming trunk, find a
register or digit receiver that can capture the incoming signals,
connect it to the line or trunk, and inform the distant end it can
start sending. In the case of a customer line, dial-tone comes from
the register, and informs the customer the system is ready to
receive his signaling. In the case of trunks, "wink start" is
generally used.
Because of the way relays have been imbedded
in the hardware and the minds of those who design it, a "wink" is a
momentary battery reversal. Recalling that reversal of the battery
on a loop trunk is the equivalent of sending an "off-hook" signal
indicating that the called party has answered, the astute reader
will recognize that the sender must know that off-hook followed by
an on-hook about 200 milliseconds later means start sending, and
that the system as a whole must recognize that a wink is not an
answer to cause charge timing to be started. This is handled by
connecting the trunk pair through directly to the sender to
originate the call, and taking the far end of the pair through to
the register or digit receiver to accept the digits, effectively
bypassing the trunk circuits in both the originating and terminating
offices. Thus wink flows from register to sender and is properly
interpreted. After sending digits, the register and sender release
control of the trunk to the trunk circuits on each end, the calling
and called parties are connected and ringing (followed by answer,
conversation, and hangup) follows.
Complex as the above is, it is fairly simple
on one-way two-wire loop-supervised trunks. The originating switch
finds an idle trunk and cuts the talking path through to a sender.
The sender makes a closure between the two wires, and the trunk
circuit at the far end recognizes the seisure. The distant office
connects the two wires of the trunk through to the register, at the
same time disconnecting the incoming trunk circuit. The register is
arranged to have its circuitry connect voltage to the two wires
backwards from the trunk; after a timed interval, the voltage
returns to normal and the wink is completed. The sender detects the
end of the wink, and sends the called number. As soon as all the
digits are sent, the sender returns control to the outgoing trunk
circuit. The register, after receiving the digits, releases control
back to the incoming trunk circuit. Since the register polarity,
once reversed, matches the original polarity of the incoming trunk
circuit, no notice of this transition of control is sent back. The
calling party is connected to the outgoing end of the trunk, and the
called party to the incoming end. Upon answer, the second reversal
flows from incoming trunk circuit to outgoing trunk circuit (instead
of register to sender), and charging can start.
Well, it wasn't actually designed by Rube
Goldberg, and it works pretty well. Except for a few minor points.
Like most trunk circuits today go via carrier, and, as a result, are
four-wire via electronic circuitry instead of two-wire using copper
conductors. And, of course, the signaling through the carrier system
is not DC voltages on wires, but tones in the voice-frequency range
going via the speech paths. And, on the trunk facility, tone on
means on-hook and tone off means off-hook. This is actually a full
two-way signaling arrangement which is cut back to one-way by the
signaling equipment that helps the trunk facility pretend it is a
pair of wires.
When two-way E&M signaling is actually used,
it has to be converted in the trunk circuit to one-way in or one-way
out for any given call, depending on direction. In No. 5 Crossbar,
relatively complex trunk circuits make this conversion. In No. 1 ESS,
with greater limitations on trunk complexity and the number of wires
available from trunk to sender or register, the M lead must be
operated directly by the common control, and the E lead read
directly. Dial pulsing for the address signaling is relatively easy
incoming, since the E lead is scanned fast enough to pick up each
pulse. Outpulsing from the trunk circuit is harder due to the need
for accuracy in controlling the shape of the pulses, but it can be
done. Fortunately, much signaling uses pairs of voice frequency
tones (MF signaling) which can come from a sender and go to a
register via the voice facility, leaving the trunk circuits to
handle only supervision and wink start.
But wink start won't work very well on
two-way circuits. Consider. It is the busy hour. A trunk is selected
for an outgoing call and is intended for cut-through to a sender. If
a sender is momentarily unavailable, there may be a slight delay,
and the far end could seize the trunk if the near end waited until
the sender was attached before seizure was sent forward. Thus
seizure on a two-way trunk must be sent from the trunk circuit and
not from the sender as on a one-way loop trunk.
Next, let us suppose that there is no delay
in getting a digit receiver. If a wink comes back to the originating
trunk circuit, the sender will never see it. Thus the trunk circuit
must remember that the wink has come and gone so that the sender can
cut loose as soon as it becomes available.
On the other hand, let us suppose there is a
delay in getting the register on the trunk. Normally, the on-hook
signal means the sender can send, while the off-hook (wink) means
hold it for a few hundred milliseconds. What is actually done can be
one of two things: the first is to have the incoming trunk circuit
return off-hook immediately upon receiving seizure, and leave this
"delay dial" signal there until the register is attached. Then the
register can return the trunk to on-hook as a "go" signal.*
[*Footnote: Note that, in the old days,
operators keying MF digits into a distant register were trained to
wait when a lamp showed the far end to be off hook, but to start
keying if the lamp was off. They had to have a "delay dial" signal
(lamp on) to stop their manual keying until the register was
attached. A 200-millisecond wink as a start dial signal was not long
enough to be seen by an operator at a busy toll board, and keying
had to be arranged to start when no light was visible. Thus "lamp
on" would delay dialing, perhaps for several seconds, while a
receiver was obtained.]
The second approach is to make the transition
from off-hook to on-hook (at the end of a wink or delay dial) tell
the sender to go. If received in the trunk circuit, prior to sender
attachment, memory of its receipt has to be provided.
Glare
Now we are in a position to consider the
final point. Suppose one switch seizes the trunk and, a few
milliseconds later, the other switch seizes the trunk from the far
end. Each is now sending off-hook toward the other; two senders are
glaring at each other. Furthermore, an off-hook is an off-hook is an
off-hook. Do we have a long wink, a delay dial, or a double seizure?
To resolve the problem, a time-out is provided at each end and,
since these timeouts are not equal, one sender will be left in
possession of the trunk and the other will get off. But here's
another problem: a sender getting off looks just like the off-hook
to on-hook transition that ends a wink or a delay dial. How can the
remaining sender tell the difference? It can't. It has to hold off
sending to see if the far end will now attach a register with a new
wink—a real one—so that digits can be outpulsed.
There are two reasons for going through all
this. First, it shows how a single problem can be made needlessly
difficult by thinking in terms of relays. Just suppose that dial
tone, like the tone a user hears when he originates a call, was
returned by any office getting a trunk seizure. Detection of dial
tone in the sender would be simple with transistors instead of
relays, economical and, most important, unambiguous. Dial tone would
be different from off-hook or on-hook. When you have dial tone, you
can send. But, alas, relays got there first, and they will still be
there when we're all retired.
The second point is closely related. All of
these problems and many others besides can be eliminated if only
some other means of signaling could be devised. Since ESS switches
are controlled by computers, and since computers talk to each other
via data links in all other businesses, why can't they do the same
thing in telephony? The answer is, they can! Thus Common Channel
Interoffice Signaling, or CCIS, was born. Hopefully, the reader can
now appreciate some of the motivation behind its development. But I
still think there is something to be said for dial tone.
Summary
The telephone network, consisting of trunking
systems, signaling systems and switching systems, united via
cross-connect frames, has grown in a curiously disconnected manner,
each component pretending to the others that it is something else.
In spite of operating, maintenance and administrative problems, the
system works. However, it is close to the limits of what it can do,
and the explosive growth in telecommunications demands changes. The
Bell System is depending on No. 4 ESS and CCIS to carry the ball.
Whether it is the right ball in the right game remains to be seen.
[
Top ] [ Table of Contents ] [
Next Chapter ] |
