Voice
Communication in Business Volume 1
Essays on telecommunications,
1969-1980
Chapter 3
Voice Communication
in Business
When I joined the RCA
Corporate Telecommunications Group, RCA had a country-wide tie-trunk
network that connected 66 locations with a total of about 33,000
extensions. Most of the equipment was Step by Step (SXS), and it
worked like a charm at a surprisingly moderate cost. As a former
Bell Labs ESS designer, I was given the task of selecting modern
interconnect PBXs to get rid of the old fashioned equipment and
effect large savings. Dick Frank, who later went to Siemens where he
designed the SD-192 PBX, joined me in this effort. During 1969, 70
and 71, we worked up a request for proposal (RFP) and sent it to
most of the major interconnect companies.
We learned a lot during
the process, and felt our researches might be of general interest.
We ultimately presented a paper based on our RFP before the
Philadelphia IEEE section, but the national IEEE expressed active
disinterest. After about a year, we sent the article to
Communications News where it was published in February, 1972.
***
A PBX is a small automatic switching system
and/or manual switchboard for interconnecting the several telephones
of a single customer (Ref. 1). A PBX differs from an intercom system
in that the telephones involved can, in addition to calling each
other, also make calls to and receive calls from the public
telephone network.
PBXs are usually relatively small. Perhaps
85% of all PBXs serve 100 lines or less. However, a study of RCA's
network shows that over half of RCA's telephones are served by PBXs
of 1,000 lines or more, and almost three-quarters by PBXs greater
than 500 lines.
Switching equipment for moderate numbers of
lines is available in three forms: PBXs, CDOs and Key Telephone
Systems. A Community Dial Office or CDO differs from a PBX by not
needing an attendant to take incoming directory number calls. A PBX
must always have someone to say "XYZ Company; Good Morning."* A key
telephone system differs from a PBX in its capacity to provide
services and features such as the ability to answer more than one
line, to hold one line while placing a call on another, to conduct
limited conferences, and, in some systems, the ability to dial
certain frequently called numbers with one or two digits. The
philosophical difference between a PBX and a key telephone system,
however, is in the nature of the traffic handled: a PBX has a high
ratio of internal to external traffic while a key telephone system's
traffic is more external than internal. Key telephone systems can be
used quite effectively without PBXs while PBXs, as they are
presently designed, almost always require the support of key
telephone systems. It is interesting to note that almost half of
RCA's monthly bill for "Service and Equipment" (excluding message
units, tolls, and tie-line costs) is for key equipment.
[* FOOTNOTE: A PBX must
also be able to transfer calls reaching wrong numbers, something
any central office switch can ignore]
There are perhaps 9 or 10 million PBX and
Centrex lines in the United States. This relatively small number of
lines originates or terminates something like 60% of all local and
long distance calls and generates about half of the total long
distance traffic. (Ref. 2,3). When one considers that somewhere
between one-third and one-half of all inter-office trunks are
dedicated to tie-line service, thus permitting PBX "internal" calls
to bypass the public network, the full magnitude of PBX operations
can be perceived.
It is our purpose here to discuss
requirements for PBXs and related equipment from the point of view
of a large business user. We hope a closer relationship between
users and designers will be encouraged thereby.
Basic PBX functions
A PBX, like any other telephone system, must
enable two or more people to talk to each other. Calls within one
PBX are easily managed with cords on a manual basis or with any one
of a variety of automatic switches. Calls internal to one company
but between users served by different PBXs can often be handled more
economically by tie lines than the public network; private tie-line
calls can be dialed directly with a minimum number of digits,
allowing desk-to-desk communication without the intervention of an
attendant or operator.
PBX connections to the public telephone
network are made via circuits called "trunks" at the PBX end and
"lines" at the central office end where, indeed, they terminate as
telephone lines. A moderate size PBX may very well have 25 or more
separate groups of CO trunks. One group, probably two-way, will
connect to the nearest CO. Several foreign exchange trunk groups are
also likely if the PBX is located in the suburbs. WATS lines,
perhaps homing on yet another CO, may also be provided. There are
six WATS bands in the United States excluding the state in which the
PBX is located, and most states offer intrastate WATS. Because WATS
circuits are one way, seven groups incoming and seven outgoing are
often needed. Further, WATS can be purchased as full time or
measured time circuits, doubling again the number of potential WATS
groups. Practically, ten groups of CO trunks would not be unusual.
Many basic PBX functions are performed by
attendants, even when Centrex service (Direct Inward Dialing coupled
with Identified Outward Dialing) is provided. Switchboard attendants
do some or all of the following: answer calls to the directory
number; locate called parties and complete the connection; transfer
calls; set up conference connections; assist internal users with
tie-line or long distance calls; keep traffic and cost accounting
records; and handle person-to-person, collect and other special
calls.
In small PBXs, the attendant often doubles as
a receptionist. She knows everybody in the company, and completes
calls by name of the called party, taking advantage of the fact that
she has just seen Mr. Brown enter Mr. Jones' office. In large PBXs,
attendants work full time at the switchboard job and usually must
have the extension number to complete. Secretaries, using key
telephone equipment, then take over some telephone attendant
functions. This becomes even more noticeable in those PBXs which are
equipped for "station dial transfer."
If a business has a number of locations in a
given area, it is often useful to centralize the PBX attendants.
Satellite systems have been used for years to distribute calls
received at a central location for remote PBXs; other techniques
connect centrally located attendants to calls received directly by
remote PBXs. When attendants are centralized only after hours, night
connections can be made using idle tie-lines. Security guards, often
using woefully inadequate equipment, may also answer calls after
hours.
PBXs provide various additional functions.
For instance, it is often necessary to deny some lines access to CO
trunks, WATS or tie lines; it is also desirable to sometimes limit
the distance calls are permitted to go without some overt means of
cost control. Specialized equipment, such as automatic call
distributors and paging systems, may be associated with PBXs in some
situations. In general, a PBX customer can be seen to have much more
complex needs than the average residential subscriber.
How step-by-step switching meets PBX needs
PBXs, satellite systems and tandem tie line
networks using SXS switches are inexpensive from less than fifty
lines to ten thousand or more, and can grow and change in a
convenient and economical manner. This comes about from the
building-block nature of the beast: each switch carries its own
control equipment which is operated directly by the user's dial.
Within a useful, but unfortunately limited, range, SXS is the
fastest and most flexible switching system ever invented. A SXS
system usually requires fewer digits than a Common Control System,
and is standing on the called line when the dial returns to normal
after the last digit. Most senderized systems, at this point, have
yet to take the first step in establishing the connection. The speed
of a non-senderized system is particularly important in tie-line
tandem switching. Because each tie-line, once seized, has its own
private selector switch on the far end, there is no delay while the
distant system scrounges up a register and attaches it. There is no
delay for "hand-shaking," and multiple dial tones are eliminated.
The only slow-down comes when a tie-line call goes off-net from the
distant PBX into the public network. Then there is a wait for the
ESS or Crossbar system to return dial tone. Fortunately, SXS systems
carry their signaling and supervision through each circuit; the
calling party can dial directly into the distant CO when it finally
notifies him it is ready.
Because each SXS switch is independent, it
can compete for circuits with all other switches in the same rank.
SXS switches can also compete with manual switchboards. All that is
necessary is to multiple the circuit to be selected past all the
equipment that might want to do the selecting and everything else
follows. Direct access to lines and trunks by both switches and
switchboards makes a SXS PBX a highly flexible system.
Another advantage of building-block
construction is the way in which the switching equipment can be
expanded selectively at any given point by simply adding switches.
For example, a company with CO trunks available by dialing 9 can add
several outgoing groups of WATS circuits by following the 9 level on
first selectors with another rank of switches; WATS bands 1 through
6 can then be reached by dialing 91, 92, and so on up to 96, with
local CO calls using 99 rather than 9.
Extra switches permit the addition of special
functions, such as segregation of CO trunks for local and tie-line
use. The incoming selectors used by tie-lines are grouped together
and reach the same station lines as the local first selectors.
However, they reach different CO trunks on the 9 level because a
separate group with a separate billing number has been multipled
past the incoming selectors only. As an additional example, consider
data switching as required by timesharing computer systems. We find
that holding times for this sort of traffic have a mean of 22
minutes and are exponentially distributed. If teletypewriters or
computer ports are put on line finders and connectors like ordinary
telephones making six minute calls with busy hour occupancy of less
than 7 ccs (about 12 minutes of use per hour), they degrade voice
service by reducing the channels available for regular calls. It is
easy to access long-holding-time terminals directly via their own
selectors so that they never delay dial tone to voice users. Data
selectors can reach data conditioned lines, MUX channels or computer
ports directly without going through switching engineered for voice
traffic only.
Adding switches is one way to achieve
flexibility in SXS systems. Adding "black boxes" such as
restrictors, message registers, TURs, trunk reversal units, and
pulse adders is another. To prevent employees from abusing telephone
privileges, restrictors can be associated with CO trunks.
Restrictors store the first several digits dialed and, if a
forbidden area or office code is detected, the call will be blocked.
Restrictors are usually positioned in such a way that dial 9 traffic
will encounter them while attendants at the switchboard can bypass
them to reach the same trunks.
Traffic usage recorders (TURs) can be used to
measure traffic on particular groups of circuits for administrative
or accounting purposes. Message registers can also be used for
similar purposes, and some large PBXs in resort hotels even have
their own automatic toll ticketing equipment paralleling that of the
central office so that guest telephone bills can be presented
instantly rather than at the end of the month. (Ref. 4). Toll
ticketing also requires "identifiers" to make available the calling
lines's identity when needed. An identifier is, of course, another
black box.
Pulse adders allow many interesting things to
be done in a SXS system, and special purpose switches capable of
"digit absorbing" and other tricks add further flexibility.
Failures of step-by-step PBXs
The good points of SXS switching also tend to
be its bad points. To do anything special, hardware at extra cost
must be added, and reasonable solutions to simple problems cannot be
achieved. For instance, one group of CO trunks could be shared by
tie-lines and local users if only suitable bookkeeping equipment
could divide up the costs of usage; restrictors would not be needed
if only the control equipment were a little smarter.
Adding switches and switchboards causes
transmission difficulties. At some points in the RCA network, as
many as 70 shelves of 10 selectors each compete for various groups
of tie-lines. The size of such a multiple amounts to a large
capacity tied across the input to the tie-line. Noise is introduced
with the use of Rotary Out-trunk Switches (ROTS) which expand the
ten outputs a SXS switch-level can reach to a maximum of 210.
Four-wire tandem switching is, in general, not available, and
several thousand 4-wire term sets fill RCA switchrooms.
Testing and maintenance are highly
unsatisfactory in SXS. Only limited self-diagnostic equipment is
built in, and most troubles must be found and corrected manually.
The system itself cannot be used to access directly specific
tie-lines for making routine transmission tests.
Class marks are very limited. Two classes are
available in most systems, and three in others; class marks must, in
general, be used on the first dialed digit if they are to be used at
all. One user class usually cannot call out of the PBX via CO trunks
and the other class has this privilege. If additional classes are
needed in connection with WATS, FX or tie-lines, little can be done.
Internal cost allocation is very difficult
with SXS PBX and tie-line systems. Although line identifiers have
been built for SXS systems in connection with Centrex, they only
work on very limited types of calls. Message registers, toll
ticketing and TURs also leave much to be desired.
Hunting groups in SXS are locked to
consecutive switch positions; this requires adjacent numbers for an
executive and a secretary, and several adjacent numbers for large
groups, implying, in turn, number changes if vacant numbers do not
naturally occur together. Further, when computer ports are to be
reached, consecutive numbers put them in the same connector group.
Ten computer ports with 22 minute mean holding times can easily
eliminate terminating call service to 90 voice users in the same
group and there is no way to balance the traffic. One must use
separate selectors as indicated above.
Finally, key telephone equipment is needed in
large quantities. This, of course, is also true of most other PBXs.
Failures of Modern PBXs
Modern PBXs offer a variety of interesting
features (Ref. 5) such as consultation hold, three way conference,
station dial transfer, call waiting, trunk answer any station, and
call forwarding ... available at an extra fee per line for all lines
on the PBX. Unfortunately, most of these features can be used only
on calls incoming from the local central office. They are usually
not available on tie-line calls, outgoing calls, or internal calls.
This comes about because intelligence in small systems is
centralized in CO trunk circuits rather than the common control. In
large systems, this particular fault is sometimes overcome, but
tie-line service then becomes almost impossible because of the
inability of modern systems to pass user dial pulses and the
inflexibility of the number of digits common control systems can
accept and act upon. In any event, even the most advanced "series
300" PBX or "Centrex II" will not be able to deal with tie-lines or
fancy gradings of WATS lines and will require the user to pay extra
for key telephone equipment which cannot be eliminated. Answer
supervision for tie-lines on off-net calls is, of course, impossible
with most non-SXS systems.
Some system deficiencies
Overall system planning for business users
leaves something to be desired. Key systems on PBXs on local COs
placing calls via tandem to other nearby COs seems particularly
wasteful. One would hope that large PBXs, at least, could home
directly on tandem offices and could be designed to incorporate key
telephone features. The advantages to the user and to the telephone
company are obvious. But most existing telephone company offerings
capable of modification to this sort of operation can't deal with
tie-lines and probably won't deal effectively with internal traffic
and cost allocation, even when the required data is generated
naturally in setting up calls.
Existing overall planning also fails on a
geographical basis. Different telephone companies have different
tariffs, and the same service varies widely not only in cost but
also in the way in which the cost is derived. Often, too, it is
impossible to obtain the same service features from one state to the
next, making uniform operation and standards impossible.
As a final example of the unsatisfactory
nature of modern telephone company offerings, consider Centrex.
Centrex consists of two service features: direct inward dialing and
identified outward dialing. It would be hard to imagine a more
blatant case of miscegenation. Direct inward dialing (Ref.6) is
highly desirable in that it reduces the need for attendants hired by
the PBX user, and saves something like 30 seconds per incoming call,
reducing traffic congestion in metropolitan areas already hopelessly
overloaded. SXS systems are particularly well suited to accept
direct inward dialing with a minimum of modification because of the
way selectors accept dial pulses immediately when seized, and
connectors return a switch-hook flash to the CO trunk circuit to
call in an attendant for assistance. Identified outward dialing, on
the other hand, is already available in most areas as QZ billing.
When automatic identified outward dialing is to be provided, at
telephone company option, it is used only on one group of CO trunks,
and then only for toll calls. Even a Centrex CO, using a No. 5
Crossbar (which generates complete information on all calls
automatically) will not provide tie-line or FX cost allocation data.
The addition of identifiers, data links, etc., in Centrex CU is a
considerable expenditure of telephone company money and user floor
space. When a user has an extensive tie-line and/or WATS network,
his use of toll may be negligible; he may even wish to use his own
attendants to control calls and costs in a variety of ways. But
there is no way to obtain DID without IOD, no matter how little the
latter may be needed.
Fortunately, the present regulatory climate
makes it possible to consider alternative approaches.
How RCA approached the problem
The RCA internal communication system needs
many new features which, in the natural course of events, do not
appear to be forthcoming. In addition, many features presently
available will pass away when SXS systems are discontinued. Finally,
many existing services and features will continue to be available
only on a "hand-crafted" basis; the need to mechanize these services
and features is of the utmost importance. Consequently, RCA is in
the process of acquiring a system suitable to its needs.
Our basic premise is this: RCA's internal
communication system should be ONE system, independent of geography,
serving telephone company, or type of electrical signals
transmitted. Standard operating procedures should apply to all
attendants and users, and the system should behave uniformly on all
calls whether incoming from or outgoing to the public network or
internal to one PBX or the entire tie-line connected system.
Further, the implemented system should
provide as much support for operating and maintenance personnel as
is reasonably possible, and information for traffic and cost
administration should be generated and stored as it occurs and
presented in a useful format at monthly intervals or as required.
Features judged necessary are: ability to
receive direct inward dialed calls and to generate internal billing
information on outward and tie-line calls; four-wire switching of
tandem tie-lines and various data stations; centralization of
key-telephone features and "Series 300" functions (Ref. 5) with
station apparatus largely limited to plain telephone sets; switched
connections to attendants over regular voice facilities; and full
use of standard telephone functions such as translation, alternate
routing, retries, uniform numbering, extensive class marking, and so
on.
The system should be sufficiently flexible to
meet future needs as they arise. This suggests a stored program
system for inexpensive changes of features, and requires a switching
matrix or equivalent which can deal with broadband data and picture
phone as well as existing low frequency signals.
The system should carry out self-testing
functions on each call as set up and on routine test calls; once
troubles are discovered, internal diagnostics should pinpoint them
for maintenance. Switching systems should also act as access devices
for specifically addressing and testing associated transmission
facilities. Moves and changes related to subsets should be greatly
simplified by translation and the removal of key-system features
from station apparatus.
The number of switchboard attendants will be
minimized, and those who remain will be able to spend their time
doing "human" rather than "machine" work. They will answer directory
number calls, help outsiders find the right party or product within
RCA, perform the internal "directory assistance" function, and
provide special call-handling assistance as required. Because calls
will be switched to them via regular facilities, attendant positions
can be located where convenient. This will permit some locations to
take over for others after hours, and allow full advantage to be
taken of the east-west time differential.
Obviously, this array of features will be
economical only for large PBXs. But such PBXs serve over half of
RCA's telephones. PBXs at smaller RCA locations, either as they
exist now or as they may be upgraded within the limits of cost
effectiveness, will home on and thus take advantage of the features
of larger systems. Larger systems will do routing and translation
and as much record keeping, routine maintenance and attendant
centralization as is practical for their smaller satellites. They
should even be able to provide direct inward dialing to select
stations on these satellites.
Equipment to perform as required will be
expensive. Nevertheless, we expect to achieve an overall saving on
communication cost as a result of the following: increased
reliability with reduced down-time and maintenance; increased
flexibility with reduction in specialized telephone instruments and
associated costs for moves and changes; optimization in the use of
attendants; minimization in the use of tie-lines through automatic
alternate routing; economical control of internal cost allocation;
and freedom from the adverse effects of future tariff variations.
By implementing our own system to our own
specifications, we have several advantages over the numerous
carriers with whom we presently deal. In the first place, we can
provide a unified communication system which can be operated and
maintained in a standard, uniform manner on a nation-wide basis.
Second, we will not have to be compatible with all possible
worst-case conditions, and all possible configurations of existing
equipment. Third, by not having to file public tariffs legally
defining the service offerings available to our users, we will have
a degree of flexibility which cannot be matched by the carriers in
offering and changing service features. Finally, by dealing directly
with our users, we can respond to their specialized needs in ways
limited only by our imagination and the capabilities of the
equipment; no limitations will be imposed by practices and
procedures hardened by man-centuries of generalized training
programs or tons of printed material sanctified by generations of
acceptance. These factors, we feel, will enable us to provide
outstanding communications to RCA for many years to come at a cost
that will remain reasonable.
References
-
IEEE Trial-Use Standard
Definitions of Terms for Communication Switching, IEEE
Transactions on Communication Technology, Vol. COM-18, No.
5, page 609, October, 1970.
-
Rodkin, D.L.,
Implementation of New Services in. PBX and Centrex Systems,
IEEE International Convention Record, Vol. 13, Part 1, page
168. 1967.
-
Brackett, G. R., PBX
Service Adviser's Handbook, Telephony Publishing Corp.,
Chicago, Ill. 1967.
-
Lopatka, E.B. and Muir,
D.W., Immediate Automatic Time and Charge Reports on Extension
Originated Toll Calls, Automatic Electric Technical Journal,
page 269, July 1967.
-
Witkus, V.W., New Services
in PBX and Centrex Systems, IEEE Transactions on
Communication Technology, Vol. COM-16, page 492, June, 1968.
-
Schleinkofer, G.N.,
Indialing to PBX Extensions: Applications in a Step-by-Step
Central Office Area, AIEE Communications and Electronics,
No. 45, page 549, Nov. 1959.
***
In retrospect, this
paper covered a lot of ground and would have been useful to any
designer reading it. In one place, however, we missed the boat. At
that time, both Dick and I hoped that single line telephones might
be able to replace key telephone sets if only the PBX had the right
features.
Northern Telecom, then
called Northern Electric (not to be confused with North Electric,
now part of ITT), asked for a copy of our RFP. As part of Bell
Canada, they had not seemed to us to be likely bidders. However, in
1975, they announced the SL-1 PBX with the best solution I have yet
seen to the key telephone problem. Maybe our RFP helped them in
arriving at this solution. I hope so.
It is interesting to
look back at a time when there were six WATS bands in the contiguous
48 states and the cost differential was enough to make them all
necessary. By the time this book is published, WATS will almost
certainly be restructured out of all recognition.
At the time, there were
no stand-alone passive recorders or active router/recorders yet on
the market.
We recognized that
adding such machines to a 701 would make the 701 cost effective
against all possible competitors for an even longer time into the
future, and weren't very eager to see them when they came along.
Some major companies, however, eventually added stand-alones very
effectively, getting the best of both worlds: telco reliability and
extra interconnected services.
It is depressing to note
that we never found a cost effective substitute for the 701 SXS
machines. Many years later, long after I had left the group, RCA
went Centrex/EPSCS. When they tore down the Camden 701, one of the
biggest still existing, I felt like I had lost one of my family.
[
Top ] [ Next ] [
Table of Contents
] |
