19.

lntroductio~i
Objectives

19.2 Stochastic Model

19.3 Classification of Stochastic Process
19.4 Specification of Stochastic ~ l o c e s s
I
19.5 Example of Markov Chain
I
19.6 Example of Markov Process with Discrete State Space
19.7 Postulates foi Poisson Process
I 19.8 Stochastic Process in Queuing

19.9 Summary
19.10 Key Words
19.1 1 Answers to SAQs

19.1 INTRODUCTION
The theory of stochastic process is now finding increasing applications to problems in
physical sciences which deal with process involving some random element in their
structure. The tendency to use the results of stochastic theory in building up of models in
applied science and engineering has increased considerably during the past decade. In the
present unit, we present an account of the general classification of stochastic processes. A
stochastic process, or sometimes random process, is the counterpart to a deterministic
process (or deterministic system) in probability theory. Instead of dealing with only one
possible 'reality' of how the process might evolve under time (as is the case, for example,
for solutions of an ordinary differential equation), in a stochastic or random process there
is some indeterminacy in its future evolution described by probability distributions. This
means that even if the initial portion (or starting point) is known, there are many
possibilities the process might go to, but some paths are more probable and others less.
Objectives
After studying this unit, you should be able to
b
explain 'stochastic process,
differentiate between various stochastic processes,
describe Markov Chain, and
define queuing model.

19.2 STOCHASTIC MODEL

"Stochastic" means being or having a random variable. A stochastic model is a tool for
estimating probability distributions of potential outcomes by allowing for random
variation in one or more inputs over time. The random variation is usually based on
fluctuations observed in historical data for a selected period using standard time-series
techniques. Distributions of potential outcomes are derived from a large number of
sin~ulations(stochastic projectiohs) which reflect the random variation in the input(s).
Its application initially started in physics (sometimes known as the Monte Car10 Method).
I It is now being applied in Engineering, Life Sciences, Social Sciences, and Finance.
Operations Research A stochastic model is one that involves probability or randomness. In this example, we
Applications have an assembly of 4 parts that make up a hinge, with a pin or bolt through the centers
of the parts. Look at the figure below, ifA + B + C is greater than D, we're going to have
a hard time putting these things together.

Figure 19.1 :A Hinge

Let's say we have a million of each of the different parts, and we randomly select the
parts we need in order to assemble the hinge. No two parts are going to be the same size!
But, if we have an idea of the range of sizes for each part, then we can simulate the
selection and assembly of the parts mathematically.
Stochastic processes (or random process or random function) are fanlilies of random
variables which are functions of say time. A few simple examples are :
Example 19.1
Consider a simple experiment like throwing a true die. Suppose that 4, is the-
outcome of the nth throw, n 2 1.Then {X,,,n L 1) constitutes a stochastic process,
also known as Bernoulli Process.
Example 19.2
Consider a random event occurring in time, such as, number of telephone calls
received at a switch board. Suppose that X(t) is the random variable which
represents the number of incoming calls in an interval (0, t) of duration t units. The
number of calls within a fixed interval of specified duration, say, one unit of time,
is a random variable X (1) and the family {X(t), t E T ) constitutes a stochastic
process (T = [0, a)).
More formally stochastic process can be defined as :
Let Q be the set of events. A stochastic process is defined as the function of two
arguments w E Q and t E T. The set T is called the parameter set or index set of the
process. It may be finite or infinite, denumerable or non-denumerable, and an interval or
a set of intervals on the real line or the whole real line R. In many physical problems, the
index t will be the time and the underlying intuitive notion will be that of a random
variable evolving in time. It is apparent that there are two ways of viewing an arbitrary
random process X (w, t ) , namely, Case (1) as a family of random variables, i.e. {(X(., t);
t E T) and Case (2) as a set of functions on T , i.e. {X(w, .) w E Q).

Thus we may regard a stochastic process either as a collection of random variables X (t)
which depend on the parameter t or as a collection of the realizations of the process X (t).
The value oEX (t) may be one dimensional, two dimensional or n-dimensional.
(When several quantities say XI (t), Xz (t) . . .X,, (t) are required for the complete
description of the state of the system at a fixed parameter point t).
For every fixed t say t = t,, X (w, t) is a random variable X(tl) = X (w, ti), the distribution Stochastic Models
of which, we denote by :

For every finite set of t-values say tl, . t, the corresponding random variables
X (1,) . . . X (t,,) have the joint n-dimensional distribution function :
.
Fn(x1,x2,..., ~ , ~ ; t...
~ t.n t) =~P r [ X (t,)_<x,... X(fn)5x,,]

One can express many essential properties of a stochastic process as properties of the
corresponding family of finite dimensional distributions.
SAQ 1
(a) Define stochastic process. What is its importance?

(b) Discuss the scope and objectives of stochastic process in industrial

engineering. I

19.3 CLASSIFICATION OF STOCHASTIC PROCESS

Classification into three classes is done by putting suitable restrictions on n-dimensional
distribution functions.
Stationary Processes
A stochastic process is stationary if its finite dimensional distributions are invariant

said to be evolutionary.
Gaussian or Normal Process
In these processes, the joint distribution function is multivariate normal. In the

Markov Process

future behavior of the process depends upon only on the present state but not on
the past. In formal terms a process is said to be Markovian, if
P r [ a , X ( t ) < b ( X , , = x , , X12 = x 2,..., X , , , ] = P r [ a < X ) ( t ) l b I X , , = x , , ]

where t I < t 2 <. . . t n < t

A discrete parameter Markov process is known as a Markov chain, 95
I
--
Operations Research
Applications 19.4 SPECIFICATION OF STOCHASTIC PROCESS
The set of possible values of an individual random variable X,, (X ( t ) )of a stochastic
process {X,,,n 1 1 ) (X (r), I E T ) is known as its state space. The state space is discrete if
it contains a finite or a denun~erableinfinity of points; otherwise it is continuous. For
example if X,, is the total number of ones appearing in the first n throws of a die, the set
of possible values of X, is the finite set of nonnegative integers 0 , 1 ,. ... Here the state
space is discrete. We can write 4, = Y , + Y2 + . . . + Y,, where Y, is a discrete random
variable denoting the outcome of the ith throw and Y, = 1 or 0, according as the ith throw
shows a one or not. Secondly consider X,, = Z, + . . . + Z,, where Z, is a continuous r . v
assuming values in [0,a).Here the set of possible values of X,, is the interval [ 0 ,m) and
so the state space of X,, is continuous.
In the above two examples we assume that the parameter TI of X,, is restricted to thz
nonnegative integers n = 0, 1 , 2 . . . We consider the state of the system at different time
points n = 0, I , only. Here the word 'time' is used in a wide sense.
In the first case considered above the "time n" implies throw number.
On the other hand, one can visualize a family of random variables {X,, t E T ) or
{ X (t), t E T ) ) such that the state of the system has characterised at every instant over a
finite or infinite interval. The system has then defined for a continuous range of times
and we say that we have a family of random variables in continuous tlme. A stochastic
process in continuous time may have either a discrete or a continuous state space. For
example suppose that X ( t ) gives the number of incoming calls at a switch board in an
interval ( 0 , t). Here the state space of X ( t ) is discrete though X ( t ) is defined for a
continuous range of time. We have a process in continuous time having a discrete state
space. Suppose X ( t ) represents the maximum temperature at a particular place in (0, t )
that the set of possible values of X ( t ) is continuous. Here we have a system in continuous
time having a continuous state space.
So far it has been assumed that the values assumed by r . v. X,, (or X (2)) are one
dimensional, but the process {$,)(or { X ( t ) ) )may be multidimensional. Consider
X ( t ) = ( X I( t ) ,X2 ( t ) )where XI represents the maximum and X2 represents minimum
temperature at a place in an interval of time ( 0 , t). We have here a two dimensional
stochastic process in continuous time having continuous state space.

Discrete Time
Continuous Value Process

Continuous Time
' Discrete Value Process

Discrete Time
Discrete Value Process

Figure 19.2 :Types of Stochastic Process

Thus a one-dimensional process can be classified into one of the following four types of
processes :
(a) Discrete time, discrete state space
(b) Discrete time, continuous state space
 (c) C O I ~ ~ ~tillle,
I ~ Udiscrete
O U ~ state space Stochastic Models

(d) Continuous time, continuous state space

All the four types may be represented by {X(t), t E T).
In case of discrete time, the parameter generally used is n, i.e. the family is represented
by {X,,, n = 0, 1,2, . . .). In case of continuous time both the symbols {X,, t E T) and
{X (t), t E T)(where T is a finite or infinite interval) are used. The parameter t is usually
interpreted as time, though it may represent such characters as d~stance,length, and so
on. Some authors call the discrete paranieter family a stochastic sequence, and the
continuous parameter family a stochastic process.
Definition
A Stochastic Process is a Counting Process N (t) if
X(t,s)=Ofort<O
X ( t , s) = integer valued and non-decreasing
+-"..,*-a .=%% ,b '

Stochastic Process Examples

Record temperature as a
- cont~nuoustime
t?.rnl
** r e
1
Record round (temperature) as
P (. r(L 6- a continuous t~rne
*i
- 1. Record temperature every
T seconds

r~ Record round (temperature)

every T seconds

Figure 19.3 : Stochastic Process Examples

Example 19.3
Consider a simple experiment that there are k boxes and an infinitely large number
of identical balls which are thrown at random one by one into the boxes. The balls
thrown being equally likely to go into any one of the boxes. If X, represents the
I
number of filled boxes after :I throws, then {X,,, tz 2 1) constitutes a stochastic
II process.
Example 19.4
Suppose that X (t) is the random variable which represents the number of
increasing calls in an interval (0,t) of duration t units. Then the family
( X (t), t E 77( constitutes a stochastic process where T = [0, co).
SAQ 2
Describe various types of stochastic processes.

19.5 EXAMPLE OF MARKOV CHAIN

Consider a simple coin tossing experiment repeated a number of times. Let us denote
head by I and tail by 0 and the random variable denoting the result of the nth toss by X,,.
Them for n = l , 2 .
Pr {X, = I) = p, Pr {X, = 0) = q
Operations Research Thus, we have a sequence of random variables XI,X2. The trials arc independent and the
Applications
result of nth trial does not depend in any way on the previous trials numbered
1 , 2 . . . (n - 1). Consider now nth random variable given by the partial sum
S,, =XI +. . . + X,.The sum S,, gives the accumulated number of heads in the first n trials
and its possible values are 0, 1, . . . n.
We have S,, + = S , + X,,, I. Given that S,, =j; (j= 0, 1, . . . n) the r . v . S,, + I can assume
only two possible values : S,,. I =j with probability q and S,, + I =j + 1 with probability
p; these probabilities are not at all affected by the values of the variables S , , . . . S,, I.
Thus, Pr { S n + ,= j + I 1 S, = j ) = p

This is an example of a Markov chain : a simple dependence that the outcome of

(n + 1) st trial depends directly on that of nth trial and only on that the conditional
,
probability of S,,, given S,, depends on the values of S,,and the manner in which the
value of S,, was reached is of no consequence.
SAQ 3
Describe Markov Chain with the help of suitable example.

19.6 EXAMPLE OF MARKOV PROCESS WITH

DISCRETE STATE SPACE
This type of process plays an important role in the study of a large number of
phenomena. One such process is Poisson process. Consider a random event E such as :
(a) incoming telephone calls (at a switch board),
(b) arrival of customers for a service(at a counter), and
(c) occurrence of accidents(at a certain place).
Consider N (t) number of occurrences of the event E in an interval of duration t, i.e. if we
start from an initial epoch (or instant) t = 0, N (t) denote the number of occurrences up to
the epoch t. For example if an event actually occurs at instants of time t , , t2, . . . then N (t)
jumps abruptly from 0 to 1 at t = tl from 1 to 2 at t = t2 and so on. The values of N (t) here
are observed values of the random variable N (t). Let p,, (t) be the probability that the
r- . v. N (t) assumes the value n.

P, (t) = Pr {N (t) = n)
This probability is a function of the time I. Since the only possible values of n are
n = 0, 1, ...

Thus {p,,(t)) represents the probability distribution of the r. v. N ( t ) for every value of I.
The family of random variables {N (t), t 2 0) is a stochastic process. Here the time t is
continuous, the state space of N (t) is discrete and integral valued. Such a process is also
called a counting process. One of the most important integral valued processes is Poisson
process, which serves as a mathematical model for a wide range of empirical phenomena
with remarkable accuracy.
 We now sho\v that undcr certain conditions, N (t) follows Poisson distribution with mean Stocl~asticModels

ht (h being a constant). In case of many empirical phenomena, these conditions are

approximately true and the corresponding stochastic process {N(t)} follows the Poisson
law.
SAQ 4
What is application of Markov Process?

1 19.7 POSTULATES FOR POISSON PROCESS

( Independence
N ( t ) is independent of the number of occurrences in an interval prior to the
interval (0, I ) i.e. future changes in N (t) are independent of the past changes.
/ Homogeneity in Time
p,,(t) depends only on the length of the interval and is independent of where this
interval is situated, i.e. p,, (t) gives the probability of the number of occurrences in
the interval (t,, t + tl) (which is length oft) for every t.
Regularity in an interval of infinitesimal length h, the probability of exactly one
occurrence is h h + o (h) and that of more than one occurrence is of o ( h ) (o (h) is
used as a symbol to denote a function of h which tends to 0 more rapidly than h.)

I In other words, if the interval between t and t + h is of short duration h, then

I q (h) = h h + 0 (h)
P, (h) = o (h) k 2 2
--
Since C P,, (h) = 1
n=O

[ Theorem
Under the postulates I, II and 111, N ( t ) follows Poisson distribution with mean h t,
i.e. is given by the Poisson law

Corollary 1
E (N (t)) = h t
and var {N(t)) =h t

The mean number of occurrences in an interval of length t is h t so that the mean

number of occurrences per unit time ( t = l), i.e. in an interval of unit length is h.
The mean rate h per unit time is known as the parameter of the Poisson process.
Postulate I implies that the Poisson process id Markovian. Postulate IT impliesthat
the Poisson process is time homogeneous, Postulate 111 implies that in an
infinitesimal interval of length h, the probability of exactly one occurrence is
approximately proportional to the length h-of that interval and that of the
simultaneous occurrence of two or more events is exactly rare.
Operatio~~s
IZesearch Poissou Process and Related Distributions
Applicatio~~s
With a Poisson process, {N(t)} the number of occurrences of an event E by epoch
t, there is associated a random variable-the interval X between two successive
occLlrrences of E.
Theorem
The interval between two successive occurrences of a Poisson process {N(t)}
1
having parameter h has a negative exponential distribution with mean - .
h
Let X be the random variable'repr&enting the interval between two successive
occurrences of {N (t)) and let Pr (I(<x) = F (x) be the distribution function

The density function is

F (x) = F' (x) = h e- " (X > 0)

Theorem
If the intervals between successive occurrences of an event E are independently
1
distributed with a common exponential distribution with mean then the events
A
E form a Poisson process with mean h t.
Brownian Motion Process --
It is an example of a continuous'parameter, continuous state space Markov
process. Markov process with continuous state space is known as diffusion
process. A particle under diffusion or undergoing Brownian motion is known as
Brownian particle.
The stochastic process {X(t), t 2 0) is called a Wiener process (or a
Wiener-Einstein process or a Brownian motion process) with drift p and variance
parameter o2if
(a) X (t) has independent increments, i.e. for every pair of disjoint
intervals of time (s, t) and (u, v) where s t u v the random variables
X (t) - X (s) and X ( v ) - X (u) are independent.
(b) Every increment X (t) - X (s) is normally distributed with mean
p ( t - s) and variance o2(t - s).
Note that (i) implies that Wiener process is a Markov process with independent
increments and (ii) implies that a wiener process is Gaussian.

19.8 STOCHASTIC PROCESS IN QUEUING

The queuing problem is identified by the presence of a group of customers who arrive
randomly to receive some service. The customer upon arrival may be attended to
immediately or may have to wait until the server is free. The service time required to
serve the customer is also a statistical variable. This methodology can b e applied in the
field of business (banks, booking counters) transportation and every day life. A queue is
formed when the current demand for a given service exceeds the capacity to provide the
ervice. In most cases. additional service facilities could be provided to decrease queues
%rtiprevent queues from forming, however the cost to provide the additional service
may cause the margin of profit to drop below an acceptable level. On the other hand
excessively long waiting lines result in lost sales and lost customers. Hence the problem
arise because of too much demand.
Too little demand means excessive idle time at service facility. The probleni faced by Stochastic hlodels
management is how to balance the cost associated with waiting against the cost
associated with the prevention of waiting in order to maxirnise profits. An analysis of
queuing systems will provide the answer to this problem under fairly general conditions.
However before looking at how queuing problems can be solved consider the general
framework of a queuing system.
A queuing system involves customers (students, airplanes, machines) arriving at a
constant or variable rate for service at a service facility. If arriving custonlers can enter
the service faculty, they do so. If they must wait for the service they join or begin a
queue, and remain in the queue until they can be served. They are then serviced at a
constant or variable rate and leave the system. The queuing system involves both the
waiting line and the service facility.
There are many types of queuing systems, but all can be classified according to the
following characteristics :
The Input or Arrival Pattern
This includes the distribution of the number of arrivals per unit of time, the
number of queues that are permitted to form, the maximum queue length, and the
maximum number of customers desiring service.
The Service Process
This includes the distribution of the time to service a customer, the number of
servers, and the arrangement of servers (in parallel, in series, etc.).
Queue Discipline
*

This is the manner in which customers form a queue (first come first served, last
come first served, random selection priority selection).
Notations and Assumptions
(a) Service is provided on a first come first served (FIFO) basis.
(b) Customers arrive completely at random but a certain average rate.
(c) The queuing system is in a steady state.
Theses three assumptions are valid in many realistic queuing systems and will
serve to illustrate the use of queuing theory.
Assumption (a) says that a customer who arrives first will be served first regardless
of whether he joins a queue or not. Assumption (b) states that an arrival is equally
likely to occur at any time and is independent of the time that has elapsed since the
last arrival. This is equivalent to saying that the number of arrivals per unit of time
is a random variable with a Poisson distribution.
That is, if
X = Number of arrivals per unit time

Then f (x) = Pr ( X = x ) = e LC x = 0, 1, 2, . .
x!

Where h is the average number of amvals per unit time? Other interesting result of
assumption (b) is that the time between consecutive arrivals T (also called the inter
arrival time) has an exponential distribution with the same parameter h.
Thus if T = time between consecutive arrivals

Then g(t)=he-" t>0

1
E ( t )= -
h
 O p e r a t i u ~ ~Research
s In summary if the number of arrivals per unit of time has a Poisson diqlribution
Applications
with mean h , then the time between consecutive arrivals has an exponential
1
distribution with mean - . The queuing system in this situation is said to have a
h
Poisson input and customers are said to arrive according to a Poisson process.
Assuniption (c) means the queuing system has been operating long enough to be
independent of the initial state of the system and is independent of time. That Is the
system has reached a state of equilibrium with respect to time. The distribution of
the number of arrivals per unit of time and the distribution of the service time do
not change with tinie.
Suppose we let
Sn = Number of custoniers in the systeni,
PI,(t) = Probability of n customers in the system at tinie t.
A,, = Average arrival rate when n custoniers are in the systeni (both
waiting and being served), and
p,, = Average service rate when n custonlers are in the system.
The system being in the steady state does not imply that the arrival rate and service
rate are independent of the number of customers in the system. For finite queue.
Limited source and multiple servers' model A,, and p,, will be functions of the
number of customers in the system. In steady state we will use the notation
P,, = probability of n customers in the system at any point of t ~ m e ,
Queuing niodels with Poisson input-exponential service.
Here we assume
(a) Arrivals to tlle system occur con~pletelyat random.
(b) Arrivals form a single queue.
(c) First in first out discipline (FIFO).
(d) Departure from the system occur completely at random.

(e) The probability of an arrival in the interval t to t + A I, for A t

sufficiently small, when the system in state S,, at time t is h,, A t.

(f) The probability of a departure in the interval t to t + A t, for

sufficiently small when the system is in state S,, at time t is p,, A t.
(g) The probability of more than one arrival and/or departure in the
interval t to t + A t for A t sufficiently small, when the system is in
state S,, at time t is 0. A t, i.e. negligible.
With these basic assumptions we can derive set of differential equations, which
when solved will yield P,, (I).

I The system can be in state Sf,at time t + A t in exactly four distinct ways :
(a) System is in state S,, at time t and no arrival or departure occur during
the interval A t.
(b) System is in state S,, - at time t and one arrival and no departure occur
during the interval A t.
(c) System is in state S, + at time t and no arrival and one departure occur
during the interval A t.
(d) One arrival and one departure or more that one arrival and/or
departure occur in the interval A t.
 Since these four ways are al nii~tuallyexclusive, the probability of n custoniei-s in Stort~asticModels
the system at time t + A t with n > 0 may be expressed as the sum of three
independent compound probabilities (the probability of way 4 above is assumed to
be negligible)
el (t + A t ) = el (t) (1 - A l r At) (1 - p I r At) + Pn-, (t) ~1,-1 At (1 - pn-1 At)

+ e,+~ (t) P,+I A ~ (-Ik I 2 +a~t )

= <,-I ('1 An-l At - Pn (1) (All + PU) At + (1) PI,+, At + C1 (I)

All higher order terms in A t are assumed to be negligible compared to A t. If P,,(t)

is subtracted from both sides we get

P, (t + At) - 5
1 (1)
At
- e,-1(1) A,,-, - e, (1) (An + P,,) + el+l(1) PI?+,+ ta-mS in At
Taking limit as A t tends to zero.
We also assume here liniit P,,(t) = P,,as t tend to infinity.
Then the rate of change of P,,with respect to time is zero.
Thus, in steady state the differential equations become the difference equations
A,,-, e,-l + P I , + , e1+1 -A,, + P n ) P n =O n>O

- A , Po + t i l 4 =O 11=O

Solving above equations

A
4 = -3(P")
111

P2 =-A" A1 p,
PI P2

p? =
P1 P2 P3

PI P2 113 . . . PI1
Using the fact that
m

rr
C
=0
el=l

lnfinite Queue - infinite Source, Single Model

Assume
Operations Research (b) The average service rate is constant; p, = p for all 11.
Applications
(c) The average arrival rate is less than the average amval rate; h < p.
This assures that an infinite queue will not be built up.
With these assumptions

For this model let

X = Number of arrivals per unit time.

The parameter h, then is the average arrival rate per unit of time. Also let
-
T Time to service a customer.
Then g ( t ) = ~ l e - ~ t' t>O; p > 0
1
E (t) = -
Cl
We already know that if the number of amvals per unit of time has a Poisson
distribution with parameter h, then the inter arrival time the time between
consecutive arrivals) has an exponential distribution with the same parameter.
Likewise if the service time has an exponential distribution with parameter p
(completely random departure) then the number of custon~ersserviced per unit of
time has a Poisson distribution with parameter p.
Thus if X, = Number of customers serviced per unit of time.
S
Then F(X,)=~-~P~---! x S = 0 , 1 , 2,...
XY
.
E (Xs) = P
One is generally interested in modeling a given queuing system and then in
studying certain interesting characteristics to determine if measures should be
taken to modify the queuing system. Modification might take the form of adding
additional servers to the service facility, providing more space in the queuing area,
or buying a new piece of equipment to increase the service rate. Most decisions are
based on several useful quantities which can be obtained through the use of
equation for Po and P,,, namely
h
L = -= average number of customers in the system
P-A
h2
Lq = = average number of customers in the queue
P (Cl - A)
h
L, =- = average number of customers in nonempty queue
P-A
 I
w=- = average time a custonlers spends in the system
P-1
h
TI = C l (11 - A) = average time a customers spends in the queue
1
w =- = average time a customers spends in the queue if he must wait
" Cl-h

P (1, > k ) = = probability of more than k customers in the system

'i ).'I/

-P\I
P(T>t)=e = probability the time in the system is greater than t

Relations among some characteristics

L=h W
L, =h W,,

Example 19.5
A person repairing mobiles finds that the time spent on the mobile set has
exponential distribution which niean 20 minutes. If the mobiles are repaired in the
order in which they came in and their arrival is approximately Poisson with an
average rate of 15 for 8-hour day. What is the repaimlan's expected idle time each
day? How many jobs are ahead of the average set just brought in?
Solution
15 1
Arrival rate h = -- = - unitslmin
8 x 6 0 32

1
Service rate p = - unitslmin
20
h 5
Exponential No. of jobs in the system = L, = -- -
p-h 3
No. of hours for which the repairman remains busy in an
8h
8-hour day = - = 5 hours
P
:. Repairman's expected idle time each day = 8 - 5 = 3 hours.
SAQ 5
(a) What effect does doubling h and 11have on L, L,/, Wand W,in a single
channel queuing model.
(b) In a large maintenance department, fitters draw parts from the parts-stores
which is at present staffed by one storeman. The maintenance foreman is
concerned about the time spent by fitters getting parts and wants to know if
the employment of a stores labourer to assist the storeman would be
worthwhile on investigation it is found that
(i) a simple queue situation exists,
 (ii) fitters cost Rs. 2.50 per hour,
(iii) the storeman costs Rs. 2 per hour and can deal, on the average, with
10 fitters per hour,
(iv) a labourer could be employed at Rs. 1.75 per hour and would increase
the service capacity of the stores to 12 per hour, and
(v) on the average 8 fitters visit the stores each hour.

Infinite Queue-Infinite Source, Multiple-Server Model

Many realistic queuing systems have more than one server providing the service in
the service facility.
We assume :
(a) S servers.
(b) Each server provides service at the same constant average rate p.
(c) The average arrival arte is constant h,, = h for all 11.
(d) h<sp.
With these assuniptions we have

Finite Queue-Infinite Source, Single Server Model

This is the situation in which the waiting line can accommodate only a finite
number of arrivals. If a customer arrives and the queue is full, the customer leaves
without joining the queue. Typical situations are :
(a) Limited space for cars that arrive at a single window drive-in bank.
(b) Limited space for cars that arrive at a do-it-yourself car wash.
(c) Limited waiting space in a one barber shop
Let M = Maximum number of customers that can get in the system at any
one time.
For this case, h need not be less than p since the queue cannot build up without
bound.
Here, p,, = p for n = 1 , 2 ,...
h , , = h for n = 1 . 2 ,... M - 1
=O for n = M , M + l ...
 -
--
I for >.=[I
.M +I

A library wants to improve its service Eacilities in terms of the waiting time of its
borrowers. 'Tl~elibrary has to counters at present and borrowers arrive according to
Poisson distribution with arrival rate 1 every 6 minutes and service time follows
exponential distribution with a meall of I0 minutes. The library I ~ a srelaxed its
men~bersl~ip rules and a substantial increase in the nun~berof borrowers is
expected. Find the number of additional counters to be provided if the arrival rate
is expected to be twice the present value and the average waiting time of the
borrowers must be limited to half the present value.
Solution
1
h = - minutes = IOIhour
6

. . New arrival
1'= 20ihour

Therefore, it is necessary to provide 2 additional counters.

 Source hlultiple Server Model
I:i~~ile
In this model, the number of custoniers desiring service is finite. Consequently, the
effective arrival ratc is a function of the service rate of each server, the nuniber of
servers. the number of customers in the callirlg population. antl the distribution of
the tinie until service is needed" for each customer. Thus we assume each customer
arrives for service according to a Poisson process at the average rate of h per hour.
This is in contrast to the total population of customers arriving at an average rate h
per hour, as the case with an infinite population. Every queuing systelii ha in fact a
finite source, but the source is generally large enough to assunie the customers
conie from in infinite source. How large, then riiust the calling population be to
justify using infinite source model? There is no single answer to this question
because it is a function of the arrival rate and the service rate.

19.9 SUMMARY
Stochastic means being or having a random variable. A stochastic model is a tool for
estimating probability distribution of potential outcomes by allowing for random
variation in one or niore inputs over time. In this unit, you have studies about types of
stochastic niodels, specification of stochastic process and applications of stochastic
models.
-
19.10 KEY WORDS
Stochastic Model : A stochastic model is one that involves probability
or randomness.
Markovian Process : A Markovian process is a time-based process
where the probability of a state on the next trial
(time period) depends only on the state in the
current trial (time period) and not on how we got
to the present state.
Markovian Chain : A Markovian chain is a stochastic process with the
following properties :
(a) Discrete state space
(b) Markovian property
(c) One-step transition probability that remain
constant over tinie (termed stationary
transition probabilities).

19.11 ANSWERS TO SAQs

v

Please refer the preceding text for all the Answers to SAQs.

FURTHER READINGS
Cooper, L. and Steinberg D., (1974), Methods arid Applications of Linear Prograrnnti~zg.
Saunders, Philadelphia, USA.
Mustafi, C. K., (1988), Operations Research Methods and Practice, Wiley Eastern
Limited, New Delhi.
Sasieni, M. A. Yaspan and L. Friedman, (1 9591, Operations Research Methods and
Problems, J . Wiley and Sons, New York, USA.
Hadly, G . , (1969), Linear Programming, Addison Wesley Reading Massachusetts, USA.
Mittal, K. V., (1976), Optimisation Methods and Operation Research and System
Analysis. Wiley Eastern Limited, New Delhi.