TELEPHONE TRAFFIC
The term "traffic," with reference to telephone service, has come to
mean the gross transaction of communication between telephone users.
This traffic may be expressed in whatever terms are found convenient for
the particular phase considered.
=Unit of Traffic.= With reference to payment for local telephone
service, the conversation is the unit of traffic. In the daily
operations of telephone systems there are fewer conversations than there
are connections and fewer connections than there are calls, because
lines are found busy and all calls to subscribers are not answered.
For these reasons, in traffic inquiries which have to do with the amount
of business which subscribers attempt to transact, the total traffic in
a given time usually is considered as so many calls originated by the
subscribers in the community. From this condition arises the term
"originating calls."
For the reason that the purpose of the switching equipment in a central
office is to make connections, the abilities of operators and of
equipments frequently are measured in terms of connections per hour or
per other unit of time.
For the reason that in charging for service all unavailing calls are
omitted, the conversation is the unit of traffic.
=Traffic Variations.= Telephone-exchange traffic is subject to such
general variations as are noted in the way a compass needle points
north, the migrations of birds, the blowing of the trade winds, and
other natural phenomena. There are variations in traffic which occur
each day, others which change with the seasons, and still others which
are related to holidays and other special commercial and social events.
For instance, the day before Thanksgiving Day, in many regions, is the
busiest telephone traffic day in the year.
[Illustration: WESTERN ELECTRIC MOTOR-GENERATOR CHARGING SET]
The daily variations in telephone traffic are closely related to
commercial activities and certain general features of this daily
variation are common to all telephone systems everywhere. Fig. 452 is a
typical graphic record of the traffic of a telephone exchange and
represents what happens in almost every town or city. The total calls in
this figure are not given as absolute units but would vary to adapt the
figure to a particular case. The figure shows principally that the
traffic in the night is light; that it rises to its maximum height
somewhere between 10 o'clock A.M. and noon; that though it is never as
high again during that day, the afternoon peak is over 80 per cent as
great; and that two minor peaks appear about the dinner hour and after
evening entertainments.
[Illustration: Fig. 452. Load Curve]
_Busy-Hour Ratio._ If the story told by Fig. 452 were to be turned into
a table of calls per hour, the busiest hour of the day would be found to
correspond to the highest portion of the figure, and in that busiest
hour of the day, if a number of selected days were to be compared, would
be found a very constant traffic. The number of calls made, or the
number of connections completed, in that particular hour, day by day,
would be found to be much the same. The ratio of the number of units in
that hour to the number of units in that entire day would be found to
be practically the same ratio day by day. This ratio of busy hour to
total day would be found to be much more nearly constant than the gross
number of calls per hour or per day.
In a large, busy city, about one-eighth of the total daily calls are in
some one hour; in a smaller, less active city, probably one-tenth are so
congested. This is reasonable when one remembers that in the larger city
the active business of the day begins later and ends earlier.
=Importance of Traffic Study.= A knowledge of the amount of traffic in
an exchange, and its distribution as to time and as to the divisions of
the exchange, is important for a number of reasons. Traffic knowledge is
essential in order that the equipment may be designed and placed in the
proper way and the total load distributed properly on that apparatus and
its operators.
For example, in an office equipped with a manual multiple switchboard,
the length of the switchboard is governed entirely by the number of
operators who must work before it. It is mechanically possible to make a
switchboard for ten thousand lines only 15 feet long, seating seven
operators. The entire multiple of ten thousand lines could appear three
times in such a switchboard. The seven operators could not handle the
traffic we know would be originated by ten thousand lines, with any
present system of charging for service. Even a rough knowledge of the
probable traffic would enable us to approximate the number of operators
needed and to equip each position, not only with access to the ten
thousand lines to be called, but also with just enough keyboard
equipment, serving as tools, and just enough answering jacks, serving as
means of bringing the traffic to her. It is foreknowledge of traffic
which enables a switchboard to fit the task it is to perform.
=Rates of Calling.= The rates of calling of different kinds of lines
vary. The lines of business stations originate more calls than do the
lines of residences. Some kinds of business originate more calls than
others. Some kinds of business have a higher rate of calling in one
season than in others. Flat-rate lines originate more calls than do
message-rate lines. When a line changes from a flat rate to a message
rate, the number of originating calls per day decreases. An operator's
position, handling message-rate lines only, can serve more lines than if
all of them were at flat rates. The number of message-rate or
coin-prepayment lines which an operator's position can care for depends
not only on the traffic but on the method of charging for service,
whether by tickets or meters and upon the kind of meters; or it depends
on the method of collecting the coins. In some regions, the rate of
calling, on the introduction of a complete measured-service plan, has
been reduced to one-fourth of what it was on the flat-rate plan.
In manual switchboards of early types, wherein the position of the
subscriber's answering jack was fixed by his telephone number, the
inequality of traffic became a serious problem. Most of the subscribers
who first installed telephones when the exchange was small, retained
their telephones and numbers; as their use of the telephone grew with
their business, it was customary to find the positions answering the
lower numbers much more busy than the positions answering the higher
numbers, the latter belonging to later and usually less active business
places.
_Functions of Intermediate Distributing Frame._ The intermediate
distributing board was invented to meet these conditions of unequal
traffic upon lines and of variations in traffic with changes of seasons
and of charges. The intermediate distributing board enables a line to
retain its number and its position in the multiple, but to keep its
answering jack and lamp signal in any desired position. If a flat-rate
subscriber changes to a message rate, his line may be moved to a
message-rate position and be answered, in company with others like it,
by an operator serving many more lines than she could serve if all of
them were flat rate.
=Methods of Traffic Study.= The best way to learn traffic facts for the
purposes of designing and operating equipment is to conduct systematic
series of observations in all exchanges; to record them in company with
all related facts; and to compare them from time to time, recording the
results of the comparisons. Then when it is required to solve a new
problem, the traffic data will enable the probable future conditions to
be known with as great exactness as is possible in studies with relation
to transportation or any other human activity.
TABLE XIII
Calling Rates
+-------------------------+-------------------------------+
| | CALLS PER DAY WITH DIFFERENT |
| KIND OF SERVICE | METHODS OF CHARGE |
| +-------------+-----------------+
| | FLAT RATE | MESSAGE RATE |
+-------------------------+-------------+-----------------+
|Residence | 8 | 4 |
|Business | 12 to 20 | 8 to 14 |
|Private Exchange Trunk | 40 | 25 |
|Hotel Exchange Trunk | 50 | 30 |
|Apartment House Trunk | 30 | 18 |
+-------------------------+-------------+-----------------+
There are three general ways of observing traffic. A record of
originating calls is known as a "peg count," because the counting
formerly was done by moving a peg from place to place in a series of
holes. The simplest exact way is to provide each operator with a small
mechanical counter, the key of which she can depress once for each call
to be counted. A second way is to determine a ratio which exists, for
the particular time and place, between the number of calls in a given
period and the average number of cord circuits in use. Knowing this
ratio, the cord circuits can be counted, the ratio applied, and the
probable total known. The third method, which is applicable to offices
having service meters on all lines, is to associate one master meter per
position or group of lines with all the meters of that position or
group, so that each time any service meter of that position is operated,
the master meter will count one unit. This method applies to either
manual or automatic equipments.
=Representative Traffic Data.= For purposes of comparison, the following
are representative facts as to certain traffic conditions.
_Calling Rates._ The number of calls originated per day by different
kinds of lines with different methods of charge are shown in Table XIII.
_Operators' Loads._ The abilities of subscribers' operators to switch
these calls depend on the type of equipment used, on the kind of
management exercised, and on the individual skill of operators. With
manual multiple equipment of the common-battery type, and good
management, the numbers of originating calls per busy hour given in
Table XIV can be handled by an average operator. The number of calls per
operator per busy hour depends upon the amount of trunking to other
offices which that operator is required to do. In a small city, for
example, where all the lines are handled by one switchboard, there is no
local switching problem except to complete the connection in the
multiple before each position. In a large city, where wire economy and
mechanical considerations compel the lines to be handled by a number of
offices with manual equipment, some portion of the total originating
load of each office must be trunked to others. Table XIV shows that an
increase of 90 per cent in the amount of out-trunking has decreased the
operator's ability to less than 70 per cent of the possible maximum.
TABLE XIV
Effect of Out-Trunking on Operator's Capacity
+----------------------------+---------------------------------------+
|PER CENT ORIGINATING CALLS | CAPACITY OF SUBSCRIBERS' OPERATOR'S |
|TRUNKED TO OTHER OFFICES | POSITION IN CALLS PER BUSY HOUR |
+----------------------------+---------------------------------------+
| 0 | 240 |
| 10 | 230 |
| 30 | 200 |
| 50 | 185 |
| 75 | 170 |
| 90 | 165 |
+----------------------------+---------------------------------------+
_Trunking Factor._ In providing the system of trunks interconnecting the
offices, whether the equipment be manual or automatic, it is essential
to know not only how much traffic originates in each office, but how
much of it will be trunked to each other office and how many trunks will
be required. An interesting phase of telephone traffic studies is that
it is possible to determine in advance the amount of traffic which can
be completed directly in the multiple of that office and how much must
be trunked elsewhere. Theoretical considerations would indicate that if
the local multiple contains one-eighth of the total lines of the city,
one-eighth of the calls originating in that office could be completed
locally and seven-eighths would be trunked out. In almost all cases,
however, it is found that more than the theoretical percentage of
originating calls are for the neighborhood of that office and can be
completed in the multiple. This results in the determination of a factor
by which the theoretical out-trunking can be multiplied to determine the
probable real out-trunking. In most cases, the ratio of actual to
theoretical out-trunking is 75 per cent, or approximately that. In
special cases, it may be far from 75 per cent.
_Trunk Efficiency._ The capacities of trunks vary with their methods of
operation and with the number of trunks in a group. For example, in the
manual system where trunk operators in distant offices are instructed
over call circuits and make disconnections in response to lamp signals,
such an incoming trunk operator can complete from 250 to 500 connections
per busy hour. The actual ability depends upon the number of distant
offices served by that operator and upon the amount of work she has to
perform on each call.
The number of messages which can be handled by one trunk in the busy
hour will depend upon the number of trunks in the group and upon the
system employed. It appears that the ability of trunks in this regard is
higher in the automatic system than in the manual system. For the
latter, Table XV gives representative facts.
TABLE XV
Messages per Trunk in Manual System
+----------------------------+------------------------+
| NUMBER OF TRUNKS IN GROUP, | MESSAGES PER TRUNK PER |
| MANUAL SYSTEM | BUSY HOUR |
+----------------------------+------------------------+
| 5 | 7 |
| 10 | 9 |
| 20 | 12 |
| 40 | 15 |
| 60 | 18 |
+----------------------------+------------------------+
Some of the reasons for the higher efficiencies of trunks in the
automatic system are not well defined, but unquestionably exist. They
have to do partly with the prompter answering observable in automatic
systems. The operation of calling being simple, a called subscriber
seems to fear that unless he answers promptly the calling party will
disconnect and perhaps may call a competitor. The introduction of
machine-ringing on automatic lines, where existing in competition with
manual ringing on manual lines, seems to encourage subscribers to answer
even more promptly. The length of conversation in automatic systems
seems to be shorter than in manual systems. Still more important,
disconnection in automatic systems is instantaneous during all hours,
whereas in manual systems it is less prompt in the busiest and least
busy hours than in the hours of intermediate congestion. The practical
results of trunk efficiencies in automatic systems are given in Table
XVI.
TABLE XVI
Messages per Trunk in Automatic System
+----------------------------+------------------------+
| NUMBER OF TRUNKS IN GROUP, | MESSAGES PER TRUNK PER |
| AUTOMATIC SYSTEM | BUSY HOUR |
+----------------------------+------------------------+
| 5 | 15 |
| 10 | 22 |
| 20 | 28 |
| 40 | 32 |
| 60 | 34 |
+----------------------------+------------------------+
_Toll Traffic._ Toll or long-distance traffic follows the general laws
of local or exchange traffic. Conversations are of greater average
length in long-distance traffic. The long-distance line is held longer
for an average conversation than is a local-exchange line. The local
trunks which connect long-distance lines with exchange lines for
conversation are held longer than are the actual long-distance trunks
between cities. Knowing the probable traffic to be brought to the
long-distance switching center by the long-distance trunks from exchange
centers, the number of trunks required may be determined by knowing the
capacity of each trunk. These trunk capacities vary with the method of
handling the traffic and they vary as do local trunks with the number of
trunks in a group. Table XVII illustrates this variation of capacity
with sizes of groups.
TABLE XVII
Messages per Trunk in Long-Distance Groups
+--------------------------+-------------------------+
| NUMBER OF LONG-DISTANCE | MESSAGES PER TRUNK PER |
| TRUNKS IN GROUP | BUSY HOUR |
+--------------------------+-------------------------+
| 5 | 2 |
| 10 | 3 |
| 20 | 3.2 |
| 40 | 3.5 |
| 60 | 4 |
| 100 | 4.6 |
+--------------------------+-------------------------+
=Quality of Service.= The quality of telephone service rendered by a
particular equipment managed in a particular way depends on a great
variety of elements. The handling of the traffic presented by patrons is
a true manufacturing problem. The quality of the service rendered
requires continuous testing in order that the management may know
whether the service is reaching the standard; whether the standard is
high enough; whether the cost of producing it can be reduced without
lowering the quality; and whether the patrons are getting from it as
much value as they might.
In manual systems, the quality of telephone service depends upon a
number of elements. The following are some principal ones:
1. Prompt answering.
2. Prompt disconnection.
3. Freedom from errors in connecting with the called line.
4. Promptness in connecting with the called line.
5. Courtesy and the use of form.
6. Freedom from failure by busy lines and failure to answer.
7. Clear enunciation.
8. Team work.
_Answering Time._ There is an interrelation between these elements. Team
work assists both answering and prompt disconnection. The quality of
telephone service can not be measured alone in terms of prompt
answering. Formerly telephone service was boasted of as being
"three-second service" if most of the originating calls were answered in
three seconds. Often such prompt answering reacts to prevent prompt
disconnecting. Patient, systematic work is required to learn the real
quality of the service.
As to answering, the clearest, truest statement concerning manual
service is found by making test calls to each position, dividing them
into groups of various numbers of whole seconds each, and comparing the
percentage of these groups to the whole number of telephones to that
position. For example, assume each of the calls to a given position to
have been answered in ten seconds or less, in which
100 per cent are answered in ten seconds or less;
80 per cent in eight seconds or less;
60 per cent in six seconds or less.
It is probable that a reasonably uniform manual service will show only a
small percentage answered in three seconds or under. Such percentages
may be drawn in the form of curves, so that at a glance one may learn
efficiency in terms of prompt answering.
_Disconnecting Time._ Prompt disconnection was improved enormously by
the introduction of relay manual boards. Just before the installation of
relay boards in New York City, the average disconnecting time was over
seventeen seconds. On the completion of an entire relay equipment, the
average disconnecting time was found to be under three seconds. The
introduction of relay manual apparatus has led subscribers to a larger
traffic and to the making of calls which succeed each other very
closely. A most important rule is, _that disconnect signals shall be
given prompt attention either by the operator who made the connection,
by an operator adjacent, or by a monitor who may be assisting_; and
another, still more important one is, _that a flashing keyboard lamp
indicating a recall shall be given precedence over all originating and
all other disconnect signals_.
_Accuracy and Promptness._ Promptness and accuracy in connecting with
the called line are vital, and yet a large percentage of errors in these
elements might exist in an exchange having a very high average speed of
answering the originating call. Indeed, it seems quite the rule that
where the effort of the management is devoted toward securing and
maintaining extreme speed of original answering, all the other elements
suffer in due proportion.
_Courtesy and Form._ It goes without saying that operators should be
courteous; but it is necessary to say it, and keep saying it in the most
effective form, in order to prevent human nature under the most
exasperating circumstances from lapsing a little from the standard,
however high. The use of form assists both the operators and the
subscribers, because in all matters of strict routine it is much easier
to secure high speed and great accuracy by making as many as possible of
the operations automatic. The use of the word "number" and other
well-accepted formalities has assisted greatly in securing speed, clear
understanding, and accurate performance. The simple expedient of
spelling numbers by repeating the figures in a detached form--as "1-2-5"
for 125--has taught subscribers the same expedient, and the percentage
of possible error is materially reduced by going one step further and
having the operator, in repeating, use always the opposite form from
that spoken by the calling subscriber.
_Busy and Don't Answer Calls._ Notwithstanding the old impression of the
public to the contrary, the operator has no control over the "busy line"
and "don't answer" situation. It is, however, of high importance that
the management should know, by the analysis of repeated and exhaustive
tests of the service, to what extent these troubles are degrading it. In
addition to improving the service by the elimination of busy reports,
there is no means of increasing revenue which is so easy and so certain
as that which comes from following up the tabulated results of busy
calls.
_Enunciation._ It must be remembered that clear enunciation for
telephone purposes is a matter wholly relative, and the ability of an
operator in this regard can be determined only by a close analysis of
many observations from the standpoint of a subscriber. A trick of speech
rather than a pleasant voice and an easy address has made the answering
ability of many an operator captivating to a group of satisfied
subscribers.
_Team Work._ By team work is meant the ability of a group of operators,
seated side by side, to work together as a unit in caring for the
service brought to them by the answering jacks within their reach. In
switchboards of the construction usual today, a call before any operator
may be answered by her, or by the operator at either the right or the
left of her position. In many exchanges this advantage is wholly
overlooked. In the period of general re-design of central-office
equipments about fourteen years ago, a switchboard was installed with
mechanical visual signals and answering-jacks on a flat-top board, and
an arrangement of operators such that the signal of any call was
extremely prominent and in easy reach of each one of four or possibly
five operators. Associated with the line signals within the reach of
such a group was an auxiliary lamp signal which would light when a call
was made by any of the lines so terminating. It was found that with this
arrangement the calls were answered in a strictly even manner, special
rushes being cared for by the joint efforts of the group rather than
serving to swamp the operator who happened to be in charge of the
particular section affected by the rush.
This principle has been tried out in so many ways that it is astonishing
that it is not recognized as being a vital one. The whole matter is
accomplished by impressing upon each operator that her duty is, _not_ to
answer the calls of a specific number of lines before her, but to
answer, with such promptness as is possible, _any call which is within
the reach of her answering equipment_.
=Observation of Service.= All that is required to be known concerning
the form of address and courtesy may be learned by a close observation
of the operators' work by the chief operators and monitors, and by the
use of listening circuits permanently connected to the operators' sets.
It is naturally necessary that the use of these listening circuits by
the chief operator or her assistants must not be known to the operators
at the times of use, even though they may know of the existence of such
facilities.
With a well-designed and properly maintained automatic equipment, the
eight elements of good manual service reduce themselves to only one or
two. Freedom from failure by busy lines and failure to answer are
service-qualities independent of the kind of switching apparatus. Too
great a percentage of busy calls for a given line indicates that the
telephone facilities for calls incoming to that subscriber are
inadequate. The best condition would be for each subscriber to have
lines enough so that none of them ever would be found busy. This is the
condition the telephone company tries to establish between its various
offices.
In manual practice it is possible to keep such records as will enable
the traffic department to know when the lines to a subscriber are
insufficient for the traffic trying to reach him. As soon as such facts
are known, they can be laid before the subscriber so that he may arrange
for additional incoming lines. In automatic practice this is not so
simple, as the source and destination of traffic in general is not so
clearly known to the traffic department. Automatic recorders of busy
calls are necessary to enable the facts to be tabulated.