Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
2~2~ ~3
-- 1 --
FIELD OF THE INVENTION
The present invention relates to a method and
a system for monitoring and forecasting customer traffic
and customer servicing at a location wherein each
customer desires to be served at any one of a plurality
of available service stations.
BACKGROUND OF ~HE INVENTION
A system and method for forecasting bank
traffic and scheduling work assignments for bank
personnél are described in Canadian patent 1,258,304
issued August 8, 1989 to Katsof et al. This patent
discloses a system which uses data gathering means for
sensing the arrival and departure of customers as well as
when a teller is at a station. Data processing means
counts the arrivals and departures and measures the
amount of time that the teller is available at each
station and the amount of time that the teller is active.
Two detectors are used: one at the entrance to a queue,
the other at the exit of the queue. The forecasting
method uses a queuing model to obtain forecast of waiting
time per customer and teller utilization. A record is
kept of the number of arrivals observed during each an
interval and of the average service time per each day of
the week.
21)2~ 5~3
OBJECT~ AND 8TATEMENT OF THE INVENTION
It is an object of the present invention to
provide a system and a method for monitoring and
forecasting customer traffic and customer servicing in a
manner which greatly reduces mathematical computations,
which increases the accuracy in real time and which uses
less components than in the systems of the prior art.
This is achieved by providing a system and a
method which uses a computing approach known as the
Discrete Event Modeling and Simulation (DEMS) which is
well known in the literature and is described in various
articles, including "Introduction to Simulation and SLAM
II", Third Edition, 1986, A. Allan B. Pritsker, chapter
11 (pages 380-427).
Discrete event modeling consists in modeling a
system by describing the changes that occur in the system
at discrete points in time. Basic to this method is the
concept of "event time" which is defined as an isolated
point in time where the state of the system may change;
an "event" is known as the associated logic for
processing the changes in state of the system.
A discrete event model of a system is
constructed by defining types of events that can occur
and then by modeling the logic associated with each event
type.
~r o ~
-- 3
The dynamic behavior of the system is produced
by representing, in a time-ordered sequence, the changes
in states according to the logic of each event.
In discrete event simulation, the system status
changes only at the beginning of an activity when
something is started or at the end of the activity when
something is terminated. Events are used to model the
start and completion of activities.
When an event occurs, the state of the model
can change in four ways:
1. by altering the value of one or more
variables;
2. by altering the m~mber of entities
present;
3. by altering the values assigned to one or
more attributes of an entity; and
4. by altering the relationship among
entities through file manipulation.
To clarify the above definitions and the
application of the DEMS approach, consider the following
event model of a single-queue single-server situation.
Consider a bank with one teller. The "states"
of the system will be measured by the number of customers
in the system and the status of the teller.
2~21~3
-- 4
The "event types" are:
1) a customer arrival event; and
2) an end-of-service event.
It will be assumed that all significant changes
in the system status can occur only at the arrival time
of a customer or at the time the service ends, in other
words, the system does not change status between these
two event times.
The activities of the system are reduced to one
only, namely the service activity of the customer. This
activity begins at either the time of arrival of the
customer or at the time that the teller completes service
for another customer. The starting time of service
activity can be either in the customer arrival event or
in the end-of-service event.
The status of the teller is described by the
variable BUSY. A value O for BUSY denotes "teller is
busy" while a value 1 denotes "teller is idle".
Customers are represented as "entities" with
one attribute denoting arrival time of the customer.
This attribute is used for collecting statistics on time
in the system for each customer.
- 5 _ ~ 5~3
The model uses a file (queue) ranked first-in,
first-out (FIF0) for storing entities (customers) waiting
for service when the teller is busy.
The state of the model at any instant of time
is defined by:
- the value of the variable BUSY;
- the location of the customer;
- the entities; and
- the attribute values of the entities.
This example shows that a complete modeling and
simulation of the bank-teller system needs only two types
of inputs, namely:
1) customer-arrival times (iOe. sensing the
customer as he or she enters the waiting
line), and
2) teller-service time (i.e. sensing the
service time of a customer by the sensor
at the teller booth).
With these two inputs, it is possible to model
and simulate the behavior of the system and collect
statistics based on observations (customer waiting time)
and time persistent variables (teller utilization).
The present invention therefore relates to a
system for monitoring and forecasting customer traffic
2 ~ 3
-- 6
and customer servicing at a location wherein each
customer desires to be served at any one of a plurality
of available service stations, the customers being in a
defined area having an entrance and an exit, the system
comprising:
A. data gathering means including:
a) detecting means, at the entrance, for
detecting passage of a customer at the
entrance;
b) display means, adjacent the exit,
activatable for displaying
identification of an available service
station;
c) sensing means, at each service
station, for sensing the presence of a
customer proximate to the station;
d) means for collecting information
relative to the detecting means,
displaying means and sensing means;
B. enabling means for allowing or disallowing
the display means to indicate the
availability of a service station; and
C. data processing means including:
- means for registering the time when
passage is detected;
~Q2~3 ~3
-- 7
- means for registering the time when
the sensing means is activated or
deactivated at each station;
- means for registering the time when
the display means is activated or
deactivated;
- means for registering the time when
the enabling means allowed or
disallowed said displaying means to
indicate the availability of a service
station; and
- means providing time correlation
between the times registered by the
registering means; this time
correlation being established by using
a discrete event modeling and
simulation method to thereby monitor
and forecast customer traffic.
The present invention also relates to a method
for monitoring and forecasting customer traffic a-nd
customer servicing at a location wherein each customer
desires to be served at any one of a plurality of
available service stations, the customers being in a
defined area having an entrance, and an exit, comprising
the steps of:
a) detecting, at the entrance, passage of a
customer thereby;
2~2~5~3
b) registering the time when passage is
detected;
c) enabling display means located adjacent
the exit to allow or disallow
identification of an available service
station to a customer;
d) registering the time when the display
means is enabled and disabled;
e) activating enabled display means located
adjacent the exit to identify, to a
customer, an available service station;
f) registering the time when the display
means is activated or deactivated;
g) sensing, at each service station, a change
of state resulting from a customer being
proximate, or not, to the station;
h) registering each time a change of state
occurs at each station; and
i) establishing time correlation between the
times registered by using a discrete event
modeling and simulation method to thereby
monitor and forecast customer traffic.
Some of the advantages obtained with the
present invention are:
1) only one detector is needed at the
entrance of the queue instead of two used
in the prior art systems;
2~2~583
g
~) the interaction between events with
reference to the real time allows to
de~ermine the presence of customers at a
station with respect to the in~ormation
displayed by the display means (direction
and teller number); it is thus possible to
determine if a customer came from other
than the waiting line;
3) time correlation between different event
times allows the self-adjustment of the
system in case of skew (or difference),
between the system waiting line and the
real one;
4) historical data may be accumulated as
usually; the resulting statistical
distributions or tables can be applied to
load forecasting by feeding them into the
simulated system using the DEMS approach.
Other ob~ects and further scope of
applicability of the present invention will become
apparent from the detailed description given hereinafter.
It should be understood, however, that this detailed
description, while indicating preferred embodiments of
the invention, is ~iven by way of illustration only,
since various changes and modifications within the spirit
and scope of the invention will become apparent to those
skilled in the art.
~ Q ~ 3~ 3
-- 10 --
IN T~E D~AWINGS:
Figure 1 is a schematic drawing of a system made in
accordance with the present invention.
DESCRIPTION OF PREFERRED ENBODI~ENTS
Referring to figure 1, there is shown a defined
area, generally denoted 10, having an entrance 12 and an
exit 14. In this area, there may be a single queue, such
as the one illustrated as 16, or various queue
arrangements allowing customers to be present at some
time at the exit 14 for servicing. Further, the
customers in this defined area may not be arranged in any
pattern at all as long as some display be used or call
made to indicate to one customer that it is time for him
or her to approach the exit 14 for servicing.
A detector 18 is located at the entrance 12 and
detects the passage of a customer in or out of the
entrance 12. Should there be more than one entrance,
additional detectors are required for each of those
entrances.
A display 20 is located adjacent the exit 14 to
inform the customer at the exit which of a series of
service stations A, C, D, E or F is available for
servicing~ A sensor 22, 24, 26, 28, 30, 32, preferably
hidden, is located at each teller station to detect the
presence of a customer in a customer detecting zone 34,
ti x ~
~, 38, 40, 42, 44 associated with each teller station.
The sensor at each station detects the arrival or
departure of a customer to that station as well as the
entrance or exit of the customer in the customer
detecting zone. In other words, if a customer being
served moves out of the detection zone for some reason
and then returns therein for further servicing, the
sensor will count two presences.
A connection box 46 interconnects the sensors
22, 24, 26, 28, 30, 32, the display 20 and the entrance
detector 18. Box 46 is connected to an input/output
processing unit 48 which forms part of a computer system
that also includes a CRT display 50 and a printer 52.
The CRT display 50 is used by an operator to
enable or disable the display 20; when enabled, the
display will provide identification as to what service
station is available. The operator also can determine
whether a teller is present, or not, at its or her
station.
In operation, customers enter into the waiting
line 16 thus activating the entrance detector 18. The
processing unit 48 registers the time at which the
activation occurs. If a customer exits from the area by
the entrance, the detector 18 will count this passage.
rS
~ 12 ~
Customers usually wait in line for service
until the display 20 indicates the identification number
of a free teller and, in most cases, the direction for
the customer to reach the available tPller.
When a customer reaches the teller station and
enters the detection zone, his or her presence is
detected by the sensor located at that station. When
this sensor changes state, the time of this occurence is
recorded. When a teller is available for servicing, his
or her number and direction are shown on the display 20
and a bip sound is activated to call the attention of the
customer located at the head of the waiting line or at
exit 14.
Preferably, there is one channel 54 connecting
to a dual display 20 with the processing unit 48. The
purpose of the dual display is to address independently
two possible directions and two numerical displays. The
system produces a delay between these two displays so
that the customer at the head of the waiting line sees
only one display (one direction with one teller number)
thus having no ambiguity as to where to go. It displays
the available stations in the same order as they were
released.
The entrance detector and the sensors are
the only devices which trigger a change of state in the
2~2~3
- 13 -
system. Therefore, a customer is considered as an entity
with one or several attributes associated to it. One of
these attributes is the arrival time, i.e., the time of
activation of the entrance detector.
The system keeps files of customer-arrival
times, customer-served times and a file representing
customers, identified with their arrival time and waiting
for service.
Real time correlation of the contents of
these file is effected by means of the discrete event
modeling and simulation method described above. With
this correlation, it is possible to accurately activate
the display without ambiguity, detect differences between
the system waiting line file and the real waiting line,
produce daily reports of customer service and teller
utilization, determine the statistical distribution of
customer arrival times and teller service times, and
produce load forecasting by system simulation.
In the present system, there are two software
components: the application software dedicated to the
real time operation of the system; and the application
software dedicated to load forecasting. The first
application is hasically a data acquisition system
tailored to the needs of the DEMS method. The second
application uses statistical formulae to derive the
~ ~ 2 r 3 l~
- 14 -
distributions of the historical data collected. A
discrete-event-based softwar~ simulator constructed on
the basis the method described above, is used to forecast
system load.
Although the invention has been described above
with respect with one specific application, i.e. bank, it
will be evident to a person skilled in the art that it
may be modified and refined in various ways to be used in
other application where a customer waits in line wishing
to be served by any one of a plurality of servicing
stations. It is therefore wished to have it understood
that the present invention should not be limited in
scope, except by the terms of the following claims.
