MANAGEMENT SCIENCE:

Queuing Theory and

Models

Queuing Theory
Queuing are the most frequently encountered problems in everyday life. For example,
queue at a cafeteria, library, bank, etc. Common to all of these cases are the arrivals of objects
requiring service and the attendant delays when the service mechanism is busy. Waiting lines

1|Page
cannot be eliminated completely, but suitable techniques can be used to reduce the waiting time
of an object in the system. A long waiting line may result in loss of customers to an organization.
Waiting time can be reduced by providing additional service facilities, but it may result in an
increase in the idle time of the service mechanism.

Queuing Theory's Origin: 1900 to 1917

The origin of queuing theory can be traced back to the early 1900s, found in a study of
the Copenhagen telephone exchange by Agner Krarup Erlang, a Danish engineer, statistician
and, mathematician. His work led to the Erlang theory of efficient networks and the field of
telephone network analysis.

The origin of modern queuing analysis lie in the growth of telephone systems in Denmark
and Norway during the early 20th century. There, telephone engineers (themselves university-
trained mathematicians) used statistical techniques to estimate capacity requirements for
automatic telephone exchanges. With little prior experience on which to base design choices,
these engineers used mathematics, especially probabilities, as an aid to their work. The
Copenhagen Telephone Company (CTC), formed in 1882, employed a thriving community of
university-trained mathematicians thanks to the efforts of its chief engineer, Johan Ludwig
Jensen, perhaps best known for “Jensen’s Inequality.” Jensen himself had studied physics,
chemistry, and mathematics at Denmark’s College of Technology. As the president of the Danish
Mathematical Society, Jensen attracted young mathematics to the CTC, among them Agner
Krarup Erlang. While studying problems of estimating telephone exchange capacity
requirements under Jensen’s direction, Erlang showed that inbound calls to a switch followed a
Poisson distribution, and that a system of lines and calls would achieve statistical equilibrium. In
1917, Erlang published “Solution of Some Problems in the Theory of Probabilities of
Significance in Automatic Telephone Exchanges,” which described three formulas used to model
call activity. Two of those, the Erlang B (or “blocking” formula) and Erlang C (the “delay” or
queueing formula; sometimes called Erlang D) are still in use today. Under the Erlang B regime,
a customer who finds all servers (telephone lines) busy, departs never to return. Under the Erlang
C regime, a customer who finds all servers busy, waits in queue until a server is available.

2|Page
 In neighboring Norway, Erlang’s contemporary Tore Olaus Engset also studied physics
and mathematics, earning a degree in 1894 from the University of Oslo. At the state-owned
Telegrafverket (today’s Telenor), Engset worked on planning for an aggressive expansion of
Telegrafverket’s phone service to all Norwegians, as mandated by an 1899 national law. As part
of that work he developed a blocking formula, and also addressed other concerns, including
customer departures, traffic variations, traffic grading, and the finite source of the customer
population. While Engset’s blocking formula found some adoption outside Telegrafverket, his
other contributions (which are described in a 1915 report) did not. Other telephone engineers
appear to have preferred Erlang’s simpler model and formulas to Engset’s both due to their being
easier to calculate as well as the former’s tendency to give results which tended to be more
conservative, that is, over-estimate switch capacity requirements.

Definition
Queuing theory is the mathematical study of the congestion and delays of waiting in line.
Queuing theory (or "queueing theory") examines every component of waiting in line to be
served, including the arrival process, service process, number of servers, number of system
places, and the number of customers—which might be people, data packets, cars, etc.

As a branch of operations research, queuing theory can help users make informed
business decisions on how to build efficient and cost-effective workflow systems. Real-life
applications of queuing theory cover a wide range of applications, such as how to provide faster
customer service, improve traffic flow, efficiently ship orders from a warehouse, and design of
telecommunications systems, from data networks to call centers.

The Usage of Queuing Theory

Queues happen when resources are limited. In fact, queues make economic sense; no
queues would equate to costly overcapacity. Queuing theory helps in the design of balanced
systems that serve customers quickly and efficiently but do not cost too much to be sustainable.
All queuing systems are broken down into the entities queuing for an activity.

3|Page
 At its most elementary level, queuing theory involves the analysis of arrivals at a facility,
such as a bank or fast food restaurant, then the service requirements of that facility, e.g., tellers or
attendants.

Queues are not necessarily a negative aspect of a business, as their absence suggests
overcapacity.

Queuing Examples
Queuing theory deals with problems which involve queuing (or waiting). Typical examples
might be:

 banks/supermarkets - waiting for service

 computers - waiting for a response
 failure situations - waiting for a failure to occur e.g. in a piece of machinery
 public transport - waiting for a train or a bus

Another Example:

For example, a 2003 paper by Stanford School of Business professor Lawrence Wein et
al. used queuing theory to analyze the potential effects of a bioterrorism attack on U.S. soil and
proposed a system to reduce wait times for medications that would decrease the number of
deaths caused by such an attack.

Queuing Model
It is a suitable model used to represent a service-oriented problem, where customers
arrive randomly to receive some service, the service time being also a random variable.

The Basic Model for Queuing Theory

The queuing theory uses a simple basic model to describe operation of systems. It
consists of the so-called service station which has one or more parallel operating similar
machines or operators, and a waiting room. The clients arrive at individual random times at the

4|Page
service station. A newly arrived customer is served, if at least one of the operating units is
available, otherwise ha has to wait in the queue.

The basic model can be varied in many ways:

 Customers are not served individually, but in groups (waiting systems with

batch operation). Application: producing batches in a production plant
 Some customers leave the system before they are served (waiting systems with time
restrictions). Application: storage of perishable goods
 Not all service devices are available for all customer (service systems with limited
accessibility). Application: production lines with dedicated machines, switching
arrangements in a telephone network.
 Some customers reject to enter the queueing system, because they the queue appears
to too long for them (queueing systems with impatient customers). Application: common
customer behavior at a post office, bank or ticket office
 A customer with a higher priority displaces customers lower priority from the service
process (service systems with priority control). Application: Express lanes in a
manufacturing process
 A customer who can not be served immediately upon his arrival, is lost (loss systems).
Application: telephone calls in a telephone network
 

The flow of incoming requests is described by a so-called renewal process. For this purpose
we imagine all requests are numbered in the in the order of their arrival. The time
span In between the arrival of the (n-1)-th and the n-th customer is called inter-arrival time. The
random variables In, n = 1, 2, ..., are assumed to be distributed stochastically independent and
identical with the distribution function Fl(x), the expected value E[I] and the variance D[I]. The

5|Page
reciprocal is called the arrival rate and indicates how many customers will arrive at the
system come in average per time unit.

The service times Sn, n = 1, 2, ..., of the successive customers also interpreted

as stochastically independent and identically distribute random variables. The distribution
function of the service times is denoted by FS(x). For the corresponding expected value and the
associated variance we use the symbols E[S] and D[S]. The reciprocal is called service rate and
indicates how many customers can be handled by the operation system per time unit in average.
If there are several parallel and similar operation equipment available, the operating rate
increased according to the number of devices.

The operation policy determines the order in which waiting customers are handled. The

following rules and terms are used:

First In, First Out (First Come, First Served). The clients are served in
FIFO (FCFS)
the order of arrivals.
Last In, First Out (Last Come, First Served). The clients are served in the
LIFO (LCFS)
reverse order of arrivals.
SIRO Selection In Random Order. The next client is chosen at random.
Non- Relative priority. Some clients will be served with a higher priority. The
preemptive operation of a client already in process will be continued before serving a
priority client with an higher priority.
Absolute priority. If a newly arrived customer has a higher priority than the
Preemptive
client which is just in operation, the service process will be stopped, so the
priority
client with the higher priority can be served immediately.
Round Robin. Each client can use the service station for a certain period of
RR time. Clients whose handling requires more time, therefore have to enter the
queue multiple times.

6|Page
 For symbolic markup of operating systems D.G. Kendall and B. W. Gnedenko have
introduced the notation

A / B / c /m

The letters A and B here define the distribution type of the inter-arrival times and the service
times. The letter c represents the number of parallel servers and m refers to the capacity of the
waiting room.

The following abbreviations are used for the distribution type:

D Deterministic distribution
M Exponential distribution
Ek Erlang distribution with parameter k (k = 1, 2, ...)
Hk Hyper-exponential distribution with parameter k (k = 1, 2, ...)
PH Phase-type distribution
G General distribution
Example:   The notation M/G/3/5 denotes an queueing system with exponentially distributed
inter-arrival times, randomly distributed service times, three parallel servers and a waiting room,
where a maximum 5 clients can wait.

The performance evaluation of queueing systems is based on the following processes:

 The number of clients in the system (Nt)t>0. This process describes how many clients are
in the system at the time t.
 The process of successive staying times (or lead times) (Vn)n in N. The random variable
Vn describes the time the n-te client will stay in the system.
For computations of the basic parameters different methods of the theory of stochastic processes
can be used. The suitability of a method depends on what types of distributions are used for the
inter-arrival and the service times and whether time-dependent or stationary values  are to be
calculated. Even the basic model of queueing theory is so complicated that it can not be solved
exactly under general distribution assumptions. However, there are approximate formulas, which
have proven themselves quite well in practice and make the stochastic operation of service

7|Page
systems transparent. By the formula of Allen-Cunnen the average number of customers in the
stationary case is:

Here    means the utilization of the systems,   the waiting probability and   and   the
coefficients of variation of the inter-arrival and service times. The formula tells us that the
number of clients in the system is larger if the load on the system and the coefficients of
variation are larger. To get a small queues, therefore you have to provide sufficient capacity or
keep the variability of the system low.

The average staying time time is obtained using the formula of Little:

where   describes the input rate at the system.

REFERENCES:

 https://www.informs.org/Explore/History-of-O.R.-Excellence/O.R.-
Methodologies/Queueing-Models
 https://www.investopedia.com/terms/q/queuing-theory.asp
 https://www.mathematik.tu-clausthal.de/en/arbeitsgruppen/stochastic-models-in-
engineering-science/public-relations/the-basic-model-of-queueing-theory/
 http://people.brunel.ac.uk/~mastjjb/jeb/or/queue.html

8|Page