Voice
Communication in Business Volume 1
Essays on telecommunications,
1969-1980
Chapter 18
The Digital Future of the
Telephone Network
Of the
best of the IEEE yearly meetings is ICC, or the International
Conference on Communications. It usually takes place in June, and I
try to go when I can to find out what various friends and
acquaintances who still have to work for a living are doing. The
digital issue was in full flower at ICC 77, and I dispatched the
following comment to Communications News where it was
published in August, 1977.
***
Sir:
ICC '77 was a great
success, particularly for sessions with the words "digital" or
"telecommunications" in their titles. When the two were combined,
people stood four or five deep at the doors, hoping to hear a word
or two, or get a glimpse of a VU-Graph.
Under the circumstances,
I conducted a very small informal poll to see how many of these
interested communications engineers were aware that digital
switching via the number 4 ESS will confront analog trunks
(long-haul carrier) for many years, while digital trunks (T carrier)
will continue to confront analog switches such as Number 1 ESS until
the turn of the century, if not longer. Few had noticed this
anomaly, and most were genuinely startled when I pointed it out.
To go one step further,
most of the independent telephone manufacturers are going ahead with
plans for digital class 5 switches with compatible remote
concentrators, PBXs, and station carrier. The Bell System, without
discernible plans for digital Class 5 or PBX switching, however, is
stressing the maintenance of existing transmission standards. In
particular, 0 loss through digital Class 5 offices is expected, with
3 dB loss in toll connecting digital trunks, and conventional VNL in
analog trunks.
Since digital Class 5
offices are four-wire by nature and must include hybrids and gain,
stability problems may result, particularly on local-local
connections. It would seem much more logical to put the 3 dB loss on
the line side, using it to mask return loss variations in the
outside plant. Although this would add 6 dB loss to local-local
connections, identical to that experienced in all-digital toll
connections, the improved stability, uniformity, and simplicity
would benefit everyone.
Maintaining existing
transmission standards, often based on step-by-step technology that
is now almost completely removed from the toll plant, is hard to
understand on any terms. The "outgoing switch to outgoing switch"
measurement technique was meaningful only when selectors, manual
operator positions, and test positions all competed at the input to
trunk circuits for access to a distant office.
In such two-wire
step-by-step systems, the 0 transmission level point (TLP) at the
outgoing local selector, or the —2 TLP at the outgoing toll
selector, made sense. In crossbar and reed-switch ESS offices, where
operators and test people access trunks via the switching matrix
itself, it makes no sense at all. Just what sense it will ever make
in digital trunk switching (when there is not only no trunk circuit
but also no channel bank and no access to any circuit on an
individual basis and no physical location of a TLP) beats the heck
out of me. It doesn't even make sense in four-wire analog switching,
although questionable economic arguments (crosspoints are supposed
to be vastly more expensive than hybrids and rebalancing for return
loss whenever a frame is added) have held four-wire switches to a
minimum.
It seems very likely
that these standards are due in part to the fragmentation within the
telephone industry—local does not speak to toll, switching does not
speak to transmission, and nobody speaks to station equipment; on
the operating level, local companies and AT&T Long Lines apparently
communicate by smoke signals, if at all, as anybody who has tried to
buy a tie-trunk that crosses state lines knows.
Under the circumstances,
let us hope that the independent manufacturers will explore some of
the transmission possibilities more fully rather than follow blindly
the AT&T standards which appear to have the result, if not the
intent, of rationalizing the Bell investment in the now obsolete
non-digital Numbers 1 and 2 ESS and the non-digital Dimension PBXs.
Although it may be true that "the system is the solution" as we are
told so often, it just might be that the system is the problem when
it comes to implementing an all-digital network.
***
The above
letter just scratched the surface. I found myself digging into the
matter, and the more I looked, the more interesting the subject
became. A friend encouraged me to do a "research report" for Probe,
a small company that had just published "The Future of AT&T," a very
expensive book that took AT&T to task for some of its real and
imagined shortcomings. Because I rather admire AT&T, even if I don't
always agree with its varied and often conflicting policies (as the
reader may have noticed), I was reluctant. However, the subject was
too detailed and technical for Business Communications Review, and I
had too much time invested to throw it away. So Probe published my
collected musings in the Fall of 1978 as
"The Digital Future of the
Telephone Network, a Study in Evolving Technology." The report
sold quite well, all over the world. Several people called me to
tell me they had read it.
Probe has
graciously permitted us to reprint the executive overview here.
Hopefully, a small percentage of readers will immediately order the
full report from Probe and help make me solvent for the next month
or so.
***
Overview
Over the past hundred
years, the telephone industry has grown from modest beginnings to
one of the most complex technological achievements mankind has ever
produced. However, because of its size and complexity, transmission,
signaling and switching have developed more or less independently of
each other, and each has gone out of its way to make the others
think it is still the way it used to be when earlier generations of
hardware froze interface designs. The coming of digital electronics,
developed largely by the computer industry, has today made almost
all of the traditional approaches to communications obsolete.
Analog transmission
principles, developed to a very high degree, require elaborate means
to adjust levels of signals at different points in switching and
transmission systems. Complex administration and maintenance
procedures have been developed to implement the analog transmission
plan and, as a result, many improvements and simplifications which
digital techniques could make possible will be difficult to realize.
Although analog
transmission advances came principally on long-distance trunks
between central offices, switching developed primarily to serve
local customers. Because of the simplifications required on a
per-customer-line basis between the telephone set and the central
office, local switching and transmission systems evolved as "2-wire"
facilities; that is, they used one transmission path for signals
going in both directions. This has led to systems using copper wire
and certain kinds of switching equipment on a scale so vast that
replacement with any other approach, no matter how desirable, will
be both difficult and costly.
The evolution of
switching and signaling shows how these systems developed, and how
they have dictated the nature of future system development. However,
the continuing growth of long-distance traffic and the driving
impact of computer electronics is forcing a change in spite of the
huge inertia that is obviously present.
An idealized system,
taking advantage of modern techniques from within and without the
telephone industry, is suggested in this study to show how both
voice and non-voice traffic might be served in the future. The
network proposed by AT&T is then analyzed to show what it can and
cannot do. Finally, the transmission requirements imposed on digital
central office switches of non-Bell manufacture are discussed to
show how they may be a significant factor in holding overall digital
development to that which can be accomplished through Bell
non-digital local central office switches.
The continued growth of
data and other non-voice communications may well increase the
fragmentation of the overall telephone network. A possible solution,
taking advantage of the concentration of non-voice communication
needs at present in large business customer systems, is outlined in
this study: PBXs, at least in the larger sizes, should be homed
directly, via digital trunks, on No. 4 ESS, the Bell System's new
digital toll switching system. This can facilitate the
implementation of digital communications where they are needed, and
permit the non-digital central offices presently used throughout the
Bell companies to serve out their useful lives handling the
predominantly analog traffic of residential customers.
***
The Probe
report had just come out when the IEEE ran a large symposium on the
digital future at Vail, Colorado. For a person who had left circuit
and system design to work directly with system users and their
problems, the meeting was like something out of Alice in Wonderland.
What follows is my trip report to my readers (both of them) in
Business Communications Review.
***
The great integrated digital whoop-de-doo
In September 1978, I
visited Vail, Colorado, where the discussions of some 300 telephone
engineers regarding "Progress toward an Integrated Digital Network"
more than compensated for the lack of natural snow that early in the
season. There were six sessions, plus a keynote address on Sunday
evening. By Wednesday noon when the busses carted us off on the two
hour ride to the Denver airport, everyone who attended felt that he
knew more about progress in digital telephony than he really wanted
to know. And, unfortunately, the progress was considerably less than
many of us had been expecting.
Not that there weren't
some surprises. GTE Automatic Electric used the conference to
announce its new line of digital central office equipment, ranging
from a 96-line station carrier system to a 100,000 line CO.
Following the approach used in the GTD PBXs, these several systems
use the same plug-in cards and design philosophy, greatly
simplifying the inventory, maintenance, training and management
requirements encountered by local telephone administrations.
However, most of the
discussion was about progress that had been planned and described
over the past several years. Sessions were organized to give
planners a chance to talk first, then manufacturers, and finally
operating telephone companies. This process permitted the same field
trials and installations to serve triple duty, amplifying somewhat
the appearance of the progress reported. The number of trials and
installations is impressive, however, and shows that the independent
telephone industry has finally found a worthy successor to
Step-by-Step switching. To be completely accurate, switching alone
is not the right word. The entire local plant is being replaced with
switches, remote switching units, and station carrier. The only real
drawback in all this is the lack of compatibility from one
manufacturer to another, forcing an operating company, once a vendor
has been selected, to stick with that vendor forever.
From the standpoint of
the business communicator, the meeting was a disappointment. To most
of us, an "integrated digital network" implies one in which voice
and non-voice services are integrated into one structure, one system
which is, indeed, the solution. The professionals in the
communications industry, however, see things quite differently. To
them, digital techniques are simply a cost effective way of handling
voice, and some expressed considerable displeasure at the very idea
that those of us who work in business communications should even
question their approach. One notable PBX designer, who should be
right in the forefront of the battle to move data on T-carrier bit
streams within and among PBXs, expressed shock at the consideration
of such ideas.
Needless to say,
customers of communication services were not represented on the
program. Representatives of one major automobile manufacturer were
present, however, and expressed privately to me some encouragement
for my questions from the floor on behalf of user needs,
particularly in the digital area.
Perhaps the most
interesting presentation, running counter to the anti-data attitudes
of most, came from Western Union. I was surprised to learn that WU
has an extensive network of T-carrier (digital) transmission systems
in most major cities, and can move digital information on a digital
medium in an efficient manner. WU described their approach as
working "from the bottom up," as opposed to the "top down" approach
in the public telephone network from digital No. 4 ESS machines to
analog local switches.
Central office loss
To me, the major
technical question at the conference dealt with attenuation of the
signal through a local (Class 5) switching system. Bell Labs and
AT&T are firmly convinced that this loss should be limited to half a
dB. It may be a coincidence, but loss through an analog (obsolete)
switching system such as Nos. 1, 2, or 3 ESS, like that through
Step-by-Step and No. 5 Crossbar, just happens to be about half a dB
(measured at 1,000 Hz). For calls between or among several local
switches, 3 dB is permitted in the loss plan for the future
all-digital system, and 6 dB will be the loss on all toll calls.
AT&T seems quite determined to impose these standards on the
independent telephone industry, even though local calls do not, of
course, enter or affect Bell's long distance network.
The problem here is that
digital PBXs and local central offices are of necessity 4-wire
machines but must meet two-wire facilities, such as lines to local
telephones or PBX extensions, on a 2-wire basis. If "impedances are
not matched" at the hybrid coils which constitute the 2/4-wire
interface, echo and, ultimately, singing can result if there is
amplification in the speech path. Traditional hybrids insert 3 dB of
loss at each end of the connection; this loss must be recovered with
amplification. Electronic hybrids contain their own amplification.
With either type of hardware, reducing net loss to less than 0.5 dB
poses serious stability problems.
In toll transmission
where amplification has been available for decades in carrier
systems (which are, of course, 4-wire facilities), the echo problem
has been solved by inserting "via net loss" in all intertoll trunks
and via net loss plus 2 dB in all toll connecting trunks. In this
way, any toll connection contains a total of via net loss plus 4 dB
from one local office to the other—a minimum amount shown by years
of human factors experiments to be required to hold echo to
tolerable levels. Via net loss increases with the round trip signal
transit time; echo which becomes more and more troublesome as this
transit time increases is, thus, attenuated proportionately.
Fortunately for all, if echo is controlled, singing also is usually
prevented. In traditional transmission systems, transmission
facilities are adjusted to provide standard gain. Via net loss and
the 2 dB additional at each end of the connection are added by
resistive attenuator pads.
In digital systems, it
is currently the vogue to use "digital" pads instead of resistive
attenuators to change levels. These pads are read-only memories that
take in the digital coding of an analog signal at one level and
return another code at a different level. It's a great idea and a
brilliant application of the new technology, but unfortunately it
has one little problem: if a data signal is inserted into the
digital bit stream directly (bypassing the A/D converters), the
digital pads convert it to garbage. To transmit digital data through
the future digital network, it will be necessary to inform the
system not only about the desired destination, but also about the
nature of the transmitted signal. And transmitting voice as well as
data over the same channel (as, for instance, for coordination) will
become very difficult. This, of course, did not bother the design
engineers in Vail.
In fairness to Bell
System thinking, their attitude is based on the cost of managing the
overall network. Bell people feel that, if loss can be permitted
anywhere, it should be left for the outside plant cables to customer
locations; a dB of loss there is worth billions compared to a dB of
loss in the central office. That is, CO loss should be reduced to
permit higher losses in the cables for minimum overall cost.
This is all very well;
it is certainly the conventional wisdom. But any rational
specification would give the desired loss between speaker's mouth
and listener's ear and let the designers divide the loss up to best
suit their technology and their customers' geography. In particular,
the Bell loss-level plan ignores the distributed switching approach
that is basic to the digital CO concept. With all digital switches
presently being described, remote concentrators and station carrier
will take the switch to the customer, reducing the length of analog
signal path to produce attenuation. Echo and stability requirements
in the future all-digital network will demand 6 dB of loss on all
toll connections, but the Bell System is decreeing that such loss
will not be present in local connections. In addition to creating
potential stability problems on local connections, this policy means
that all local calls will be 6 dB louder than all toll calls, a
noticeable difference that will become more and more annoying as the
ratio of toll to local calling continues to climb.
The REA, which finances
small telephone systems in rural areas, pointed out that
specifications for their borrowers permit 2 dB central office loss
on local calls. Both the REA speaker and others described how remote
concentrators and station carrier, by eliminating long loops, are
improving transmission in spite of the 2 dB loss in the CO, and are
reducing outside plant costs at the same time.
What has all this to do
with the business customer? Just this. A PBX is, in many ways,
exactly like a remote concentrator and, in a digital context, it can
connect directly to a digital switch via a T-carrier line. This
saves cable, of course, but it also maintains digital integrity for
non-voice services over a larger area. It could be a major step
toward allowing PBXs (with the proper line cards) to send data
directly (without modems) within the entire area served by the
switch on which such PBXs home. Assuming such COs will home, via
digital trunks, on the digital No. 4 ESS, an even wider area of
digital integrity becomes possible. Imagine sending a page of
printed or typed material in less than three seconds, and some of
the possibilities are seen to be intriguing to say the least.
To show how firmly the
Bell 0-loss is fixed in the minds of the faithful, one floor
question from a Bell man actually asked if the cost comparison
between traditional cables and electronics had considered use of
finer gauge wire rather than remote concentrators; it appears that
the existing plant should be pulled out and replaced, not with
modern technology, but with cables containing less copper. Since the
main cost in outside plant is in the labor of putting the stuff up
in the first place and administering all the cross connections and
junction boxes, etc., once the cables are in operation, the very
suggestion that electronics on existing cables is more expensive
than putting up new fine gauge wire is absurd. That this kind of
thinking is limiting the future integrated digital network should be
evident.
Handling data
This brings us to the
frequently expressed attitudes of telephone people concerning data.
They repeated over and over again the litany that data requirements
are "not well defined" and thus should presumably be excluded from
planning in the "integrated" digital network of the future. (Picturephone
development, of course, continues, since it seems to fill a
well-defined need in the minds of some.)
It is possible and even
likely that Bell's proposed Advanced Communications Service (ACS)
will be able to handle data quite well for the next 50 years or so
while the analog (obsolete) ESS machines are being depreciated. And
Bell Labs people went out of their way to explain to me that they
can, indeed, do just about anything with electrical communication if
only somebody could define the problem. They even hinted darkly
about great developments in progress. But maintaining digital
integrity to the PBX and, even better, to the user telephone, does
not seem to agree with their party line.
There were a lot of
other topics covered. Digital microwave, digital satellites and
fiber optics appear to be coming along very rapidly, for instance.
And considerable effort is being applied to the problems of
depreciation and the related regulatory problems in a rapidly
changing technology. But the business customer, whose toll calls
during the day generate the profit that makes the entire telephone
industry possible, is getting the short end of the stick. If AT&T
has its way, we'll have POTS forever, using transistors and, maybe,
a few integrated circuits, but definitely reed relays which are, by
some magic at Western Electric, far less expensive than electronic
devices.
Common channel signaling
There will be a few new
features, of course. Common Channel Interoffice Signaling (CCIS) is
supposed to speed up calling (except to (a) the local central
offices on each end of a connection and (b) to the PBXs served by
such offices). Further, CCIS is supposed to make possible automatic
call-back on a nationwide basis (except for connections to lines in
hunt groups where automatic call-back doesn't or shouldn't work).
Obviously, CCIS will have to get to the local COs and then to the
PBXs before it can do all this, even if it can find a line in a PBX
that is not in a hunt group.
But of course CCIS
appears to be on the edge of obsolescence. Canada is not going to
use the AT&T version (which has a relatively low-speed, 2 Kbps,
channel). They are going directly to CCITT No. 7 or something quite
similar, and will use a 64 Kbps channel in a T-carrier line to
provide much greater capability. It is possible that the United
States can ultimately end up as an analog island surrounded by a
digital world.
Obsolete systems
Let me make one final
point. During the discussions of depreciation, several speakers and
participants from the floor noted how rapidly PBXs are becoming
obsolete, and how they provide special problems in depreciation and
regulation. An example was given concerning a hospital which, just
four years ago, had installed a modern PBX with "all the modern
features." To the lament of the speaker, this PBX was now obsolete.
I contend that this
example shows more clearly than anything else at the entire meeting
why the digital future will be slow in arriving. Systems are
designed primarily to permit the designers to demonstrate their
virtuosity in manipulating modern components. Consideration or even
knowledge of user requirements is secondary if it exists at all.
This is shown very clearly in PBX and Centrex design where the
majority of telephone planners and designers seriously believed that
"Series 300 features" could replace key telephone operations; that
users would be glad to memorize 15 or 20 "feature codes" to make the
system perform preposterous feats dreamed up in ivory towers; and
that automatic message accounting was one thing that could safely be
left off a PBX to make design easy and keep costs down.
The older PBXs are
obsolete, but they were not made obsolete by new technology. They
were obsolete before they ever came off the drawing boards. This is
readily demonstrated by the way it is easier to displace a five year
old PBX as a 20 year old Step-by-Step system. I contend that, if
telephone equipment were designed on the basis of user needs in the
first place, it would not become "obsolete" when somebody introduces
a new "feature." A properly designed PBX could only become obsolete
when new technology reduced costs so much that scrapping the old PBX
imposed no financial hardship since the new equipment would save
money over retaining the old. Obsolescence by correcting dumb
designs is a luxury that even the United States, the richest country
in history, can ill afford.
Conclusion
Well, maybe things are
different internal to the telephone network. Maybe the customer
doesn't need to know how signals are carried from here to there.
Maybe it is none of his business. Maybe, as many telephone designers
think, he is too dumb to understand the problem or to appreciate the
solution. But somehow, I don't think so. Although telephone users in
general, and business customers in particular, may not know all the
technical details, they have some pretty clear ideas about what they
want to do, and they will be very quick to think up new ways to
utilize modern technology if it can reach them at the right price.
An integrated
voice/non-voice network, preferably extending to the subsets, (at
least in PBXs), would seem to be the way to go. That we are not
about to go that way was made abundantly clear at Vail. But let's
hang in there. The game isn't over. And remember, it's your game. If
you don't let people know what you want, you will never get it. I,
for one, do not think your data needs are as "undefined" as the
telephone industry says. And you have a lot of opportunities now to
"vote with your bucks." Carterfone proved once and for all that
somebody else could not be prevented by law from providing a useful
service that the utilities did not choose to offer. But in the case
of the integrated digital network, it would be highly desirable if
somehow the needs of the users could be integrated into the evolving
telco design.
***
For those
who want to read the official meeting report, see the IEEE
Communications Society Magazine for November, 1978 (Vol. 16, No. 6),
where Amos Joel speaks for the group as a whole. For another
minority report, my good friend Ray Kraus can be depended upon. His
feelings, as usual, are diametrically opposed to mine, and can be
found in some detail in Telephone Engineer & Management, June 1,
1979, page 93.
As a
final shot, it should be noted that the regulators have a lot to say
about how the telephone network should be operated. Their background
for making such decisions is limited but, as we have seen, some of
the best technical minds in the world have come up with some very
strange things.
All this
is brought out very well in a debate conducted by the IEEE between
Walter Hinchman, former head of the FCC Common Carrier Bureau and J.
R. Pierce, retired from Bell Labs: see the IEEE Spectrum, Dec.,
1979. I couldn't help but respond and, for once, the IEEE published
my letter in the May, 1980, issue. Naturally, they left out the
technical parts that their membership wouldn't have understood, but
what remained covered the ground.
***
The "debate" between
Walter Hinchman and John Pierce on the future of the telephone
industry would be funny if it weren't so sad. Mr. Hinchman mouths
the standard platitudes of the many FCC petitioners who feel that
they have a God-given right to exploit the communication customer
because competition, per se, must be subsidized no matter how much
it costs or how badly it degrades service. Mr. Pierce, on the other
hand, follows the telephone party line by insisting that the utility
is doing such a great job that it should be allowed to continue
doing whatever if feels like in the future, independent of customer
needs. Like the blind men with the elephant, both are presenting
views that can most charitably be described as half-truths.
It should be noted that
Mr. Hinchman's friends have saddled the country with several analog
"specialized" common carriers to handle the data explosion which,
presumably, will be digital. Because experienced customers often
prefer to stay with Bell after trying their circuits in private data
networks, the specialized common carriers are now selling most of
their circuits as ordinary voice tie-trunks or as part of
competitive cut-rate voice networks operating in a cream-skimming
context.
Mr. Hinchman's statement
that "Most of these networks interconnect satisfactorily with the
local exchange networks..." is obviously equivalent to a testimonial
for the healthful properties of tobacco based on knowing a
three-pack-a-day smoker who does not yet have cancer. And so far,
Mr. Hinchman's suspicion that innovation may develop among the
specialized common carriers has proved to be unfounded (four-wire
switching and automatic message accounting can hardly be considered
innovative, even though AT&T refused for years to offer them in any
useful manner).
As for Mr. Pierce's
friends, their major fault is their desire to concentrate
development efforts on residential telephone service, which is
already pretty good, at the expense of business communication which
is, in a word, awful. AT&T's resistance to facsimile and data
transmission is a matter of record, its unwillingness to permit
foreign devices to be connected to its sacred network or to allow
resale of service is legendary, and its fragmentation of the
telephone industry along political boundaries is a matter of despair
among communication managers of interstate businesses. If AT&T were
as responsive to business customer needs as Mr. Pierce says,
Carterfone could never have happened.
Note, too, that Mr.
Pierce's hopes for an all digital network in the future will be
dashed by the long continuing existence of Nos. 1, 2 and 3 ESS, the
Dimension PBX, various subscriber concentrator and carrier systems,
TSPS and other current Bell Labs marvels which are, alas, not
digital in any respect (although they are computer controlled). On
the other hand, a completely digital toll network, based on the
addition of long-haul fiber optics to existing short-haul T Carrier
and No. 4 ESS systems, will doubtless be blocked by Mr. Hinchmans'
friends to protect the investment of the specialized common carriers
with their analog microwave and satellite systems.
It seems clear that Mr.
Hinchman knows little about the management of communication systems,
and Mr. Pierce knows even less about how business communication
managers experience the Bell System. With two such knowledgeable
gentlemen arguing the highly partisan positions of competitive
vendor groups before legislators and regulators more interested in
doctrinal purity than effectiveness or cost of telecommunication
services, the plight of the customer can hardly be viewed with
optimism.
***
Again, I
got to play prophet, but I can't take credit for it. I did a seminar
on the Digital Future with Probe in New York in January, 1980, and
the very next day, AT&T announced its intention to run a fiber
optics transmission system from Boston to Washington, stringing
together on a digital thread a series of digital No. 4 ESS beads.
And almost immediately, an analog "specialized" common carrier
intervened. But my letter wasn't printed until May, long after my
predictions were old hat.
[
Top ] [ Next ] [
Table of Contents
] |
