Escolar Documentos
Profissional Documentos
Cultura Documentos
?
m In Modeling real-world phenomena there are few
situations where the actions of the entities within
the system under study can be completely
predicted in advance
m The world the model builder sees is probabilistic
rather than deterministic
m Some statistical model might well describe the
variations.
m An appropriate model can be developed by
sampling the phenomenon of interest. Then
through educated guesses, the model builder
would select a known distribution form, make an
estimate of parameter of this distribution, and
then test to see how a good fit has been obtained
Àeview of Terminology and Concepts
Discrete random variables 3
Jet X be a random variable. If the no:of possible values of
X is finite, or countably infinite, X is called a discrete
random variable. The possible values of X may be listed as
x1, x2, ««..
E.g:
The no. of jobs arriving each week at a job shop is
observed; random variable X (no. of jobs arriving each
week)
The possible values of X are given by the range space of
X, which is denoted by Àx. Here Àx = {0,1,2««}
iscrete Àandom variable«««.
2. £ p(xi) = 1
è
Continuous Àandom Variables
a is a continuous random variable if its range space is an
interval or a collection of intervals.
·
Continuous random variable«..
ÿ
Cumulative istribution Dunction
Cumulative istribution Dunction (cdf) is denoted
by D(x), where it measures the probability that
the random variable X assumes a value less than
or equal to x, i.e.
m D() = P(X )
m If X is discrete, then
D()=Ȉall xi x p(xi)
m If X is continuous, then
m D() = - f(t) dt
ü
m Properties of cdf are,
m a) D is a nondecreasing function.
If a< b, then D(a) D(b).
b) lim xĺ D(x)=1
c) lim xĺ- D(x)=0
Expectation
m If X is a random variable, the expected value of a is
denoted by a
m If a is discrete, ¦
m If a is continuous, ¦
x f(x dx
m ¦
if X is continuous.
m a a
a
m Also, a a
a
?
Ô
m In the discrete case, the mode is the value of the
random variable that occurs most frequently.
?
Queueing Systems
m In queueing systems, the time between arrivals and
service times are always probabilistic; However it
is possible to have a constant interarrival time or
constant service time.
m Eg: Jine moving at a constant speed in the
assembly of an automobile
?è
Inventory systems
m There are 3 Àandom variables
o The number of units demanded per order or per time
period
o The time between demands
o The Jead time (time between placing an order for
stocking the inventory system and the receipt of that
order)
?·
Sample statistical models for lead time distribution :
Gamma
Sample statistical models for demand distribution:
m Poisson: simple, extensively tabulated and well
known.
m Negative binomial distribution: longer tail than
Poisson (more large demands).
m Geometric: special case of negative binomial ,has
its mode at unity ,given that at least one demand
has occurred.
?ÿ
Àeliability and maintainability
> Time to failure has been modeled with numerous
distributions, including the exponential, gamma
and Weibull.
> If only random failures occur, the time-to-failure
distribution may be modeled as exponential.
> Gamma: In a case where each component has an
exponential time to failure.
> When there are a number of components in a
system and failure is due to the most serious of a
large number of defects, or possible defects, the
Weilbull distribution seems to do particularly well
as a model .
?ü
J
m In many instances simulations begin before data
collection has been completed
?
Jimited data
?
iscrete istributions
m iscrete random variables are used to describe
random phenomena in which only integer values
can occur.
m It contains
m Bernoulli trials and Bernoulli distribution
m Binomial distribution
m Geometric distribution
m Poisson distribution
> Bernoulli Trials and Bernoulli istribution [iscrete..]
> Bernoulli Trials:
> Consider an experiment consisting of n trials, each can be
a success or a failure.
> Jet a
> and a
> xj = 1, j=1,2,«n
> pj (xj) =p(xj) = 1-p = q, xj = 0, j=1,2,«n
?
m The Bernoulli distribution (one trial):
m where mean a
m a
!
Bernoulli process:
The n Bernoulli trials are called a Bernoulli process if
i) the trials are independent
m ii)Each trial has only two possible outcomes ie success or
failure
m iii)and the probability of a success remains constant from trial
to trial.
m Thus p(x1,x2,«, xn) = p1(x1). p2(x2) «..pn(xn)
Binomial istribution [iscrete ««..]
m The random variable X denotes the no of
success in n Bernoulli trials has a binomial distribution given
by P(x) where
m p(x) = ( ) p q, x = 0,1,2,«n
It determines the probability of a particular outcome with all
the success ,each denoted by S, occurring in the first x trials
followed by the n-x failures, each denoted by an D.
m x
m
è
m The event { X=x} occurs when there are x-1
failures followed by a success.
m Probability of failure = q = 1-p
m Probability of success = p
m P(DDD...DS) = q p
'(') *
The probability of two or more beeps in a 1-hour
period:
+,-
ü
X
m V
m
m
m X
m
X
m D
%
&
m '
%
×
X V
m $
"
m ×
"
$
(((!
m V"
× )
* +×
* +V
×
* +,
* +X X
*+V
-V
!
×
×
m
!
m !
$
m
!
m /
&0
×
,
m
/
#
m ,
&
m
#
,
m
,
/1
,
%
12
!
!
m
3*+
!
m
/ $
&
(!
,
X ' X
V
V
m V
V1V7V(((
m 5V1V7((8
#
m
/
V
V
V
V
Queuing Notation :-
{
{ {
{
!
"
# $
# %$
!
$
%$
%
%
%
#
#%
%
Jong-run Measures of performance of Queuing system
Ô
1. p
3
Consider, a queuing system over a period of time T. Jet J(t)
denote the number of customers in the system at time t. Jet Ti
denote the total time during [0,T] in which the system contained
exactly i customers. In General,
i=0ü p p
The
number in the system is defined
by
J = (1/ p i=0ü i Ti = i=0ü i { Ti / T }«««««.(1)
i=0ü i Ti = 0 TJ(t) dt
è
p
Consider, a queuing system over a period of time T.
Jet Wi be the time spent by customer i in the system during
[0, T]. The average time spent in the system per customer,
called the average system time, is given by
ŵ = (1/ ´ i=1üN Wi
Where N is the number of arrivals in [0, T].
Dor stable systems as N ŵ w
è
3. The Conservation equation : J = Ȝw
J = Ȝ w This relation holds for almost all queueing
systems.
When T and N The equation becomes J = Ȝw.
It says that the average number of customers in the system at
an arbitrary point in time is equal to the average number of
arrivals per time unit , times the average time spent in the
system.
·
{
{ {
{
!
"
# $
# %$
!
$
%$
%
%
%
#
#%
%
·?
Steady ±State Behavior of Infinite population
Markovian Models
Dor the infinite population models , the arrivals are assumed to
follow a Poisson process with rate { arrivals per time unit. i.e
inter arrival times are assumed to be exponentially distributed
with mean 1/ {.
Service times may be exponentially distributed.
The queue discipline will be DID. Because of the exponential
distributional assumptions on the arrival process ,these models
are called Markovian Models.
A Queueing system is said to be in
statistical equilibrium or steadystate ,provided the probability that
the system is in a state which does not depend on time.
i.e P(J(t) = n) = Pn(t) = Pn is independent of time t.
The steady state parameter J ,the time average
number of customers in the system is
J = n=0ü n Pn
Where Pn are the steady state probabilities.
The other parameters are
w=J/{
wQ = w ± (1 / )
JQ = { wQ
Where { is the arrival rate and is the service rate
per server.
1. Single-server Queues with poisson arrivals and Înlimited
capacity: M/ G / 1( i.e capacity & calling population are infinite)
Steady state parameters are
6{/
J = '{2(1/ 2 + ı2 ) = '2 (1+ 2 ı2)
2(1- ) 2(1- )
w = 1 + {(1/ 2 + ı2 )
2(1- )
wQ = {(1/ 2 + ı2 )
2(1- )
JQ = {2(1/ 2 + ı2 ) = 2 (1+ 2 ı2)
2(1- ) 2(1- )
P0 = 1-
2. M/ M / 1 Queue : Steady state parameters are
+ 6{/
ii) J = { =
-{ (1- )
iii) w = 1 = 1
-{ (1- )
iv) wQ = { =
( -{) (1- )
v)JQ = {2 = 2
( -{) (1- )
··
2. Multiserver Queue: M/ M / c / /
Here c channels (servers) operating in parallel. These channels
has an independent and identical exponential service time
distribution with mean 1/ .The arrival process is poisson
with rate {. Arrivals will join a single queue and enter the
first available service channel. If the number in the system
is n < c , an arrival will enter an available channel .When n
c, a queue is formed.
Steady state parameters are:
{ È c
ii)P0 = {[ £
n=0
c-1 ( {/ )n] + [ ({/ )c (1 / c!) (c ) ] }-1
n! (c - {)
= {[ £
n=0
c-1 ( c )n] + [ (c )c (1 / c!) 1 ] }-1
n! (1- )
·ÿ
iii) P(J( ) c) = ( {/ )c P0 = ( c )c P0
c!(1 - {/c ) c!(1 -
iv)J = c ' (c)c+1 P0 = c ' P(J( ) c)
c (c!)(1- )2 (1 -
v) w = J
{
vi) wQ = w ± (1/ )
viii) J ± JQ = c
( {/ )c P0
c!(1 - ·ü
Steady ±State Behavior of Dinite population Models
(M/M/c/K/K)
Dor the finite population models with K customers,
the arrivals are assumed to be exponentially
distributed with mean 1/ { time units.
Service times may be exponentially distributed with
mean 1/ time units. There exist c parallel servers
and system capacity is K.
m Steady state parameters are:
i) = {e È c = (J - JQ) /c
ii)P0 = [ n=0 £c-1 (K n) ( {/ )n + n=c £K K! ( {/ )n ] -1
(K - n)! c! cn ± c
iii) Pn = (K n) ( {/ )n P0 , n = 0,1,«..c-1
K! ( {/ )n P0 , , n = c , c+1,««.K
m (K - n)! c! cn ± c
iv) J = n=0 £K n Pn
v) JQ = n=c+1 £K (n - c )Pn
vi) {e = n=0 £K (K - n) { Pn
vii) w = J
{e
viii) wQ = JQ
m {e