MA8402 - PROBABILY AND QUEUEING THEORY

CLASS NOTES
UNIT – IV QUEUEING MODELS
Queueing Theory: Queueing theory refers to the mathematical study of the formation,
function, and congestion of waiting lines, or queues.
Any queuing situation involves two parts:
 Someone or something that requests a service—usually referred to as the
customer, job, or request.
 Someone or something that completes or delivers the services—usually referred
to as the server.
Examples:
 When looking at the queuing situation at a bank, the customers are people
seeking to deposit or withdraw money, and the servers are the bank tellers.
 When looking at the queuing situation of a printer, the customers are the
requests that have been sent to the printer, and the server is the printer.

Queueing theory scrutinizes the entire system of waiting in line,

including elements like the customer arrival rate, number of servers, number of
customers, capacity of the waiting area, average service completion time, and queueing
discipline.
 Queueing discipline refers to the rules of the queue, for example
whether it behaves based on a principle of first-in-first-out, last-in-first-out, prioritized,
or serve-in-random-order.

Kendall notation:
Queuing theory uses the Kendall notation to classify the different types of queuing
systems, or nodes. Queuing nodes are classified using the notation A/S/c/K/N/D where:
 A is the arrival process
 S is the mathematical distribution of the service time
 c is the number of servers
 K is the capacity of the queue, omitted if unlimited
 N is the number of possible customers, omitted if unlimited
 D is the queuing discipline, assumed first-in-first-out if omitted
For example, think of an ATM.
It can serve: one customer at a time; in a first-in-first-out order; with a randomly-
distributed arrival process and service distribution time; unlimited queue capacity; and
unlimited number of possible customers.
Queuing theory would describe this system as a M/M/1 queue (“M” here stands
for Markovian, a statistical process to describe randomness).
Queue Characteristics:
 balking (customers deciding not to join the queue if it is too long)
 reneging (customers leave the queue if they have waited too long for service)
 jockeying (customers switch between queues if they think they will get served
faster by so doing)
 a queue of finite capacity or (effectively) of infinite capacity
Importance of Queueing theory:
Waiting in line is a part of everyday life because as a process it has several
important functions. Queues are a fair and essential way of dealing with the flow of
customers when there are limited resources. Negative outcomes arise if a queue process
isn’t established to deal with overcapacity.
For example, when too many visitors navigate to a website, the website will slow
and crash if it doesn’t have a way to change the speed at which it processes requests or a
way to queue visitors.
Queuing theory is important because it helps describe features of the queue, like
average wait time, and provides the tools for optimizing queues. From a business sense,
queuing theory informs the construction of efficient and cost-effective workflow systems.
Applications of Queueing theory:
 Queuing theory is powerful because the ubiquity of queue situations means there
are countless and diverse applications of queuing theory.
Queuing theory has been applied to:
 telecommunications
 transportation
 logistics
 finance
 emergency services
 computing
 industrial engineering, etc.,
BASIC QUEUEING PRINCIPLES AND MODELS:
Most of queueing theory deals with system performance in steady-state. That is,
most queueing models assume that the system has been operating with the same arrival
rate, average service time and other characteristics for a sufficiently long time that the
probabilistic behavior of performance measures such as queue length and customer
delay is independent of when the system is observed. Clearly, there are many service
systems, including health care systems, for which there are time-of-day, day-of-week or
seasonality affects.
Little’s Law:
Little’s Law connects the capacity of a queuing system, the average time spent in
the system, and the average arrival rate into the system without knowing any other
features of the queue.
Little’s Queueing Formula:
In many queues, it is useful to determine various waiting times and queue sizes for
Particular components of the system in order to make judgments about how the system
should be run.
Let us define L to be the average number of customers in the queue at any given
moment of time assuming that the steady-state has been reached. We can break that
down into Lq, the average number of customers waiting in the queue, and Ls, the average
number of customers in service. Since customers in the system can only either be in the
queue or in service, it goes to show that L  Lq  Ls .
Likewise, we can define W as the average time a customer spends in the queuing
system. Wq is the average amount of time spent in the queue itself and Ws is the average
amount of time spent in service. As was the similar case before, W  Wq  Ws .
It should be noted that all of the averages in the above definitions are the steady-
state averages. Defining λ as the arrival rate into the system, that is, the number of
customers arriving the system per unit of time, it can be shown that
 L  W
Lq  Wq
Ls  Ws
Markovian Birth - Death process & Steady State probabilities.
Markovian Birth - Death process:
Let n be the size of the customer at time t and let Pn (t ) be the
probability of n customers in the system at t.
Let n  average arrival rate when n customers are in the system.
n  average service rate when n customers are in the system
Now Pn (t  t ) is the probability of n customers at time t  t
The presence of n customers in the system at time t  t can happen in any one of the
following four mutually exclusive ways:
i) presence of n customers at time t and no arrival or no departure during t time.
ii) presence of n-1 customers at time t and one arrival and no departure during t time.
iii) presence of n+1 customers at time t and no arrival and one departure during t time.
iv) presence of n customers at time t and one arrival and one departure during t time .
Pn (t  t )  Pn (t )(1  n t )(1  n t )  Pn 1 (t )n 1t (1  n 1t )
 Pn 1 (t )(1  n 1t ) n 1t  Pn (t ) n t n t
Pn (t  t )  Pn (t ) 1  n t  n t  n n  t    Pn 1 (t ) n 1t  n 1n 1  t   
2 2

   
Pn 1 (t )  n 1t  n 1n 1  t    Pn (t ) n  n  t 
2 2

 
Pn (t  t )  Pn (t ) 1  (n  n )t  n n  t    Pn 1 (t ) n 1t  n 1n 1  t   
2 2

   
Pn 1 (t )  n 1t  n 1n 1  t    Pn (t ) n n  t 
2 2

 
on omitting terms containing (t )2 which is negligibly small.
 Pn (t  t )  Pn (t )  (n  n ) Pn (t )t  n 1Pn 1 (t )t  n 1Pn 1 (t )t
Pn (t  t )  Pn (t )  (n  n ) Pn (t )t  n 1Pn 1 (t )t  n 1Pn 1 (t )t ,
Pn (t  t )  Pn (t )
 n 1 Pn 1 (t )  (n  n ) Pn (t )  n 1 Pn 1 (t )
t
Taking limits on both sides of (1) as t  0, we have

Pn (t )  n1Pn1 (t )  (n  n ) Pn (t )  n1Pn1 (t )    (1)

Equation (1) holds for n  1,
When n = 0.
P0 (t  t )  P0 (t )(1  0 t )  P1 (t )(1  1t ) 1t
P0 (t  t )  P0 (t )  0 tP0 (t )  1tP1 (t )  11  t  P1 (t )
2

On omitting  t 
2

P0 (t  t )  P0 (t )  0 tP0 (t )  1tP1 (t )

P0 (t  t )  P0 (t )
 1P1 (t )  0 P0 (t )
t
as t  0,

P0 (t )  0 P0 (t )  1 P1 (t )    (2)

Equations (1) and (2) are the equations of the birth and death process.
Now in steady state, Pn (t ) and P0 (t ) are independent of time and hence Pn (t ) and P0(t )
become zero.
Hence the differential equation (1) and (2) reduce to the difference equations
n1Pn1  (n  n ) Pn  n1Pn1  0      (3) and
0 P0  1P1  0      (4)
To find steady state probabilities:
From eqn (3) derived above, we have
0
P1  P0      (5)
1
Putting n=1 in (3)
0 P0  (1  1 ) P1  2 P2  0
2 P2  (1  1 ) P1  0 P0
0
 (1  1 ) P  P
1 0 0 0
  P  10 P0  10 P0
 1 0 0
1
01
2 P2  P0
1
01
P2  P
12 0
Successively putting n=1,2,3,.... in (3) and proceeding similarly, we can get
012
P3  P etc.
12 3 0
 012 ....n1
Finally Pn  P0 , n  1, 2,3,...    (6)
12 3 ....n
Since the number of customers in the system can be 0 or 1 or 2 or 3 etc., which events are

mutually exclusive and exhaustive, we have P
n 0
n 1


012 ....n1
i.e., P0   P0  1
n 1 12 3 ....n

1
P0       (7)

   ....n 1
1  0 1 2
n 1 1 2 3 .... n

Equations (6) and (7) provide the steady state probabilities for n customers in the

system.
MODEL I
(M/M/1):(  /FCFS) Model:
The M/M/1 system is made of a Poisson arrival, one exponential (Poisson) server, First
come, first serve, queue of unlimited capacity and unlimited customer population.

M/M/1 queueing node

Ls , Lq in (M/M/1):(  /FCFS):
Derivation:
The arrival rate and service rate are constant .
To find steady state probabilities , Put n   , n   in (6)
 n
n 
Pn  n P0    P0
 
1 1
where P0  n
 1 2


 
1    1        ...
n 1     
1 
  1
 
1

1   
 
n
  
 Pn    1  
  

Average No.of customers in the system Ls:


Ls  E ( N )   nPn
n 0
n n

      
  n   1    1    n  
n 1          n 1   
             
1 2 3 2
   
 1      2    3    ...  1    1  2    3    ...
                 
2 1 1
                
 1    1    1       
                

Ls 
 

Average No.of customers in the system Lq:


Lq  Ls 

       2
  
    (   )
2
Lq 
 (   )

1. Traffic intensity = Busy period  

5. Average waiting for a customer in non-empty queue

8. Probability that there is nobody in the system


P0  1 

9. Probability that the waiting time of a customer in the system exceeds t is

(  )t
P(Ws  t)  e
10. Probability that the waiting time of a customer in the queue exceeds t is
  (   ) t
P(Wq  t )  e

11. Probability that there will be k or more customers waiting for service is
 k

p(n  k )   


PROBLEMS
1 Customers arrive at the express checkout lane in a supermarket in a Poisson process with
a rate of 15 per hour. The time to check out a customer is an exponential random variable
with mean of 2 minutes. Find the average number of customers present. What is the
expected waiting time for a customer in the system?
Solution:
1
( M / M /1) : ( / FIFO)   15 / hr &   2 min  *60  30 / hr
2
 15
Ls   1
   30  15
L 1
Ws  s  hour  4 min
 15
2 People arrive at a theatre ticket booth in Poisson distributed arrival rate of 25 per hour.
Average service time is 2 minutes following exponential distribution. Calculate (1) the
mean number in waiting line (2) the mean waiting time (3) the utilization factor.
1  25
Soln.   25 per hour ,    60  30 per hour ,     0.833
2  30
 25
Ls   5
  30  25
 25 30  5
(1) Length of queue Lq  Ls   5   4.16
 30 6
Lq
(2) Mean waiting time   9.6 min

(3) Utilisation factor    0.833

3 Customers arrive at a one window drive –in bank according a Poisson distribution with
mean 10 per hour. Service time per customer is exponential with mean 5 minutes. The
space in front of the window including that for a serviced car can accommodate a maximum
of three cars. Others can wait outside this space.
(i) What is the probability that an arriving customer can drive directly to the space in front
of the window?
(ii) What is the probability that an arriving customer will have to wait outside the
indicated space?
(iii) How long is an arriving customer expected to wait before being served?
(iv) Expected number of customers waiting in the system.
(v) Expected waiting time in the system.
Solution:
This problem under (M / M /1) : ( / FIFO)
Given Arrival rate   10 per hr
1 1
Service rate   / min     60  12 / hr
5 5
 10
Ls   5
   12  10
 10
Lq  Ls   5   4.16
 12
i) Probability that an arriving customer can drive directly to the space infront of the
window  P0  P1  P2
2 n
         
 1    1       1   Pn    1  
         
2
 10  10  10   10   10 
 1    1      1    0.42
 12  12  12   12   12 
ii) Probability that the arriving customer has to wait outside the indicated space
= 1-0.42=0.58
iii) An arriving customer expected to wait before being served
= Average waiting time in the queue
Lq 4.16
Wq    0.417 hrs
 10
iv) Expected number of customers waiting in the system
Ls  5
v) Expected waiting time in the system
L 5 1
Ws  s  
 10 2
4 Customers arrive at a one-man barber shop according to Poisson process with a mean
inter arrival time of 12 mins. Customers spend on average of 10 mins in the barber's
chair.
(i) What is the expected number of customers in the barber shop and in the queue?
(ii) Calculate the percentage of time an arrival can walk straight into the barber's chair
without having to wait.
(iii) How much time can a customer expect to spend in the barber's shop?
(iv) What is the probability that the waiting time in the system is greater than 30 mins?
1 1
This is an infinite queueing model with single server   / min = 5/hr. and   / min = 6/hr.
12 10

 5  5 25
(a) Ls    5, Lq  Ls   5  
 65  6 6
 5 1
(b) P0  1   1 
 6 6
Percentage of time an arrival can walk straight into the
1
barber's chair without having to wait = *100 =16.67%.
6
LS 5
(c) Ws    1hr.
 5
 1      12 1

(d) P Ws  30 min   P  Ws  hr   e e 2
 2 
5 Customers arrive at a one man barber shop according to a poisson process with a
mean inter arrival time of 20 min. Customers spend an average of 15 min. in the barber’s
chair, then (a) What is the probability that a customer need not wait for a hair cut ?
(b) What is the expected no. of customers in the barber shop and in the queue?
(c) How much time can a customer expect to spend in the barber shop?
(d) Find the average time that the customers spend in the queue.
(e) What is the probability that there will be 6 or more customers waiting for service?
1 1
This is an infinite queueing model with single server   / min. and   / min.
20 15
1
 1
(a) P0  1   1  20  ,
 1 4
15
 1  3 9
(b) Ls   20
 3, Lq  Ls   3    2.25
 1 1
15 20  4 4
  Ls  3 & Lq  2.25
Ls 3
(c) Ws    60 min.
 1
20
Lq 2.25
(d) Wq    45 min.
 1
20
6
    20 
6 1
(e) p(n  6)        0.1779
    115 

6 Arrivals at a telephone booth are considered to be Poisson with an average time of 12 min.
between one arrival and the next. The length of a phone call is assumed to be distributed
exponentially with mean 4 min.
(a) Find the average number of persons waiting in the system.
(b) What is the probability that a person arriving at the booth will have to wait in the
queue?
(c) What is the probability that it will take him more than 10 min. altogether to wait for the
phone and complete his call?
(d) Estimate the fraction of the day the phone will be in use.
This is an infinite queueing model with single server
1 1
Mean inter arrival time   12 min , therefore Arrival rate   / min.
 12
1 1
Mean service time   4 min , therefore Service rate  = / min.
 4

1
 12
(a) Ls    0.5 customer
  1 1

4 12
1
 1
(b) P (system is busy) =    12 
 1 3
4
(  )10 5
(c) P(Ws > 10) = e =e 3  0.1889

1
(d) P (Phone will be in use) 
3
7 In a cinema theatre people arrive to purchase tickets at the average rate of 6 per minute
and it takes 7.5 seconds on the average to purchase a ticket. If a person arrives just 2
minutes before the picture starts and it takes exactly 1.5 minutes to reach the correct seat
after purchasing the ticket.
i) Can he expect to be seated for the start of the picture?
ii) what is the probability that he will be seated when the film starts?
iii) How early must be arrive in order to be 99% sure of being seated for the start of the
picture?
Solution:
Model identification:
since there is only one counter and the arrival of persons are infinite, capacity of the
system is infinie. Hence this problem comes under the model (M/M/1);(  /FCFS).
Given Data:
Arrival rate   6 per min
1 60
Service rate   per sec     8 per min
7.5 7.5
 6  6 18
Ls    3, Lq  Ls   3    2.25
 86  8 8
i) Expected total time required to purchase the ticket and to reach the seat
=waiting time in the system + time to reach the seat = Ws  1.5 .
Ls 3 1
Where Ws     0.5
 6 2
Expected total time required to purchase the ticket and to reach the seat =0.5+1.5=2 min
ii) P(he will be the seated for the start of the picture)
=P(Total time < 2 min)
 1  1
1
 (86)
 P  Ws    1  P Ws    1  e (   )t  1  e 2
 0.63
 2   2 
iii) Suppose t minutes be the time of arrival so that he is seated 99%, then
P(Ws  t )  0.99
 1  P(Ws  t )  0.99
 P(Ws  t )  1  0.99  0.01

e(   )t  0.01  e2t  0.01  2t  log(0.01)  t  2.3

This is the waiting time in the system. that is to purchase ticket.
He takes 1.5 minutes to reach the seat after purchasing ticket.
 total time = 2.3+1.5=3.8
Hence he must arrive atleast 3.8 minutes earlier so as to be 99% sure of seeing the start
of the film.
8 A telephone exchange receives one call every 4 minutes and connects one call every 3
minutes. If the rate of arrival follows Poisson distribution and the service rate follows
exponential distribution find the average waiting time for a call in the queue and in the
system.
Model identification:
since there are one server and the capacity of the system is infinity. Hence this problem
comes under the model (M/M/1);(  /FCFS).
Given Data:
1
Arrival rate   per min
4
1
Service rate   per min
3
1 1
1 4 3 9
Ls   *12  3, Lq  3  4  3    2.25
1 1 4 1 4 4

3 4 3
i) The average waiting time for a call in the System
Ls 3
Ws    12 min
 1
4
ii) The average waiting time for a call in the queue
Lq 2.25
Wq    9 min
 1
4
9 An airport has a single runway. Airplanes have been found to arrive at the rate of 15 per
hour. It is estimated that each landing takes 3 minutes. Assuming a Poisson process for
arrivals and an exponential distribution for landing times. Find the expected number of
airplanes waiting to land, expected waiting time. What is the probability that the waiting
will be more than 5 minutes?
Solution:
Model identification:
since there is only one single runway and the capacity of the system is infinity. Hence this
problem comes under the model (M/M/1);(  /FCFS).
Given Data:
Arrival rate   15 / hr
1
Service rate   3min  *60  20 / hr
3
  15  15 9
Ls    3, Lq  Ls   3    2.25
   20  15  20 4
i) The expected number of airplanes waiting to land is
Lq  2.25
ii) Expected waiting time of the airplanes is
Lq 2.25
Wq    0.15
 15
iii) Probability that the waiting time of a customer in the queue exceeds t is
  (   ) t
P(Wq  t )  e

1 1 1
1/ 4  3  4  3  12 
P(Wq  5)  e  e  1.226
1/ 3 4
10 A TV repairman finds that the time spend on his job has an exponential distribution with
mean 30 minutes. If he repair sets in the order in which they came in and if the arrival of
sets is approximately Poisson with an average rate of 10 per 8 hour day.
a) Find the repairman’s expected idle time each day?
b) How many jobs are ahead of average set just brought?
Solution:
Model identification:
since there is only one repair man and the capacity of the system is infinity. Hence this
problem comes under the model (M/M/1);(  /FCFS).
Given Data:
Arrival rate   10 per 8hr day
1 1
Service rate  30    per min
 30
 (i.e)   8  2  16 per 8hr day
 10 6 3
i) The repairman’s idle time = P0  1   1   / day
 16 16 8
3
The Expected idle time = 8   3 hrs
8
 10
ii) The average number of jobs in the system = Ls    1.667 2 jobs.
  16  10
Another method:
10 5
Arrival rate    per hr
8 4
 1
Service rate   per min
30
60
(i.e)    2 / hr
30
 10 6 3
i) The repairman’s idle time = 1   1  
 16 16 8
3
The Expected idle time each day = 8   3 hrs
8
ii) Number of the jobs ahead of the
average set brought in = The average number of jobs in the system
5
 4
= Ls    1.667 2 jobs.
  2
5
4
MODEL - II
M / M / S : FCFS /  / 
System size probabilities for an M / M / S : FCFS /  /  queueing system under steady
state conditions.
Assume S  c   M / M / c  ;( / FCFS )
This model represents a queueing system with poisson arrivels, exponential service time,
multiple servers, infinite capacity and FCFS queue service from a single queue.
If n  c, then only n of the c servers will be busy and others are idle and hence mean service
rate will be n .
If n  c, all c servers will be busy and hence the mean service rate is c 
n , 0  n  c
n   and n   n
c  , n  c
By birth and death process
0 12 3 ....n1
Pn  P
1 2 3 4 ....n 0
Case i: When 0  n  c
.... (n times)
Pn  P0
1 2 3 4....n 

1 
Pn    P0
n!   
Case ii: when n  c
 ......(n times)
Pn  P
1 2 ....c ....c 1 ....n 0
n
 P
1 2 3...(c  1)c c...(c  (c  1)) times 0
n
 P0
(c  1)!  c 1c n c 1  n c 1
n
 P0
(c  1)! c  c 1c n c  c 1   ( c 1)
n
 P0
(c  1)!  c 1c n c 1  n c 1
n
1 
Pn    P0 , n  c
c!cn c   

 1   n
 n c   P0 , n  c
 c!c   
Pn   n
1 
 n !    P0 ,0  n  c
  
To find P0

Since P
n0
n 1

c 1 
  Pn   Pn  1
n 0 n c

n n
1 
c 1 
1 
    P0     P0  1
n 0 n !    n c n !   

 c 1 1   n 
1 
n

 P0          1 ---------(1)
 n 0 n !    
n c
n  c c !c 
Consider
n n

1  1 
  
      
n c c !c n  c  c !c  c n c   c 

1   c   c 1   c  2 
  
       ...
c !c  c   c    c   c  

1   
c
   1   2   3 
   1         ...
c !c  c   c     c    c    c  
 1
      
c
1
   1   
c !c  c   c     c 
c 1
1     c   
    
c !c  c  c   c 
n c

1  1     c 
       
n c c !c n  c  c !c  c  c   c   
1
 c 1 1   n 
1  
n

(1)  P0         
 n 0 n !       
n c
n  c c !c

1
 c 1 1   n 1      c 
c

 P0         
 n 0 n !    c !c  c   c    c    
1
 c 1 1   n 1   c   c  
 P0          
 n 0 n !    c !      c    

Expression for average number of customers in the system and waiting time of a
customer in the system:
To find the average number of customer in the system:  Ls 

Ls  Lq 

Where
Lq  Average number of customer in the queue

Lq   (n  c) Pn
n c

n

1 
Lq   (n  c)   P0
nc c !c n c 
n
1 
  
 P0  (n  c)  
 c 
c
c !c n c

1    
c 1
  
c2
  
c 3

 P  0  ( c  1  c )    ( c  2  c )    ( c  3  c )    ...
c !c  c   c   c   c 
0


1   
c 1
   
1
  
2

 c 0 
P   1  (2)    (3)    ...
c !c  c    c   c  
c 1 2
1      
 P0   1   c 
 c 
c
c !c  
 c 1

c 1 2  
Lq 
1     1 
 
c c!    1   c  P0  c c ! P
2 0
    
1   c 
 
c 1

 
 Ls  Lq 


1  P0 

 c c!   
2

1   c 
 

PROBLEMS
1. A supermarket has 2 girls running up sales at the counters. If the service time for each
customer is exponential with mean 4 minutes and if people arrive in Poisson fashion at the
rate of 10 per hour, find the following:
(i) What is the probability of having to wait for service?
(ii) What is the expected percentage of idle time for each girl?
(iii) What is the expected length of customer’s waiting time?
(iv) What is the expected number of idle girls at any time?
Solution:
Model identification: Since there are two girls and infinite capacity.
Hence this problem comes under the model (M/M/c);(  /FCFS).
Given data:
Arrival rate   10 per hr
1 60
Service rate   per min (i.e)    15 per hr.
4 4
Number of servers c  2
1
 c 1 1   n 1   c   c  
w.k.t  P0          
 n 0 n !    c !      c    
1
 1 1  10 n 1  10 2 2 15 
P0        
 n 0 n !  15  2!  15  ((2 15)  10) 
1
 1 1 30 2
   0.6667    0.6667  
n

 n 0 n ! 40 
1
 1 1 
   0.6667   0.3333
n

 n 0 n ! 
P0  (1  0.6667)  0.3333  21  0.5
1

i) The probability that a customer has to wait for the service is

 c
 10 
2

   
 15   
P( N s  2)  (0.5) P( N s  c)  P0
 10    
2!1   c !1  
 15  2   c 

 0.6667 
2
1
P( N s  2)  (0.5)   0.1667
1.3333 6
 10 1
ii) Probability of time that a girl is busy     
c 2  15 3
 Probability of time when a girl is idle = 1- Probability of time that a girl is busy
1 2
= 1 
3 3
2
 Percentage of idle time of for each girl   100  67%
3
Lq 1
iii) Expected waiting time of customer = Ws  
 
c 1 2
1    
Where Lq    1   P0
c.c !     c 
2

 0.6667  1   (0.5)  0.083

1 3 10

(2)2!  30 
Lq 1 0.083 1
Ws      0.0083 hrs
  10 15
iv) The expected idle no of girl:
E(idle no of girl)=?

No.of idle girls: 2 1 0

Probability P0 P1 P2

E(idle time for each girl)=2 P0 +1 P1 +0 P2

Now, P0 =0.5
n
1 
w.k.t Pn    P0 , 0  n  c
n!   
1
1  10  1
P1    (0.5)  0.333 
1!  15  3
 2
1  10  1
P2    (0.5)  0.1111 
2!  15  9
1 1 1 4
E(idle no of girl)=2 P0 +1 P1 +0 P2 = 2   1  0  1  
2 3 3 3
2 Four counters are being run on the frontiers of the country to check the passports of the
tourists. The tourists choose a counter at random. If the arrival at the frontier is Poisson
at the rate λ and the service time is exponential with parameter , find the average

queue length at each counter.

Solution:
( M / M / c) : ( / FIFO)

  ,   & c4
2
  1
2 & 
 c 2
1
 c 1 1    n 1   c   c  
P0          
 n 0 n !    c !      c    

 
1
 41   (4)  
 1 n 1
   2   2
4 2 
 n 0 n !

4!
 2  
  (4)    

 
1
  
 3 1 1 4 1 
   2    2  
n

 n 0 n ! 4! 1 
 1  
  2  
1
 4 8 1 
 1  2    16  2  
 2! 3! 4! 
3

23
c 1
1  1
Lq    P
c !c      
2 0

1  c 
 
1 1 3 4
 2
5
 2

4!4  1  23 23
1  
 2
3 A tax-consulting firm has 3 counters in its office to receive people who have problems
concerning their income, wealth and sales taxes. On the average 48 persons arrive in an
 8-hr day. Each tax advisor spends 15 mins on the average on an arrival. If the arrivals are
Poisson distributed and service times are according to exponential distribution, find
(i) the average number of customers in the system.
(ii)the average number of customers waiting to be serviced
(iii)the average time a customer spends in the system.
Solution:
48 1
No. of servers c = 3, λ = = 6/hr, µ = *60 =4/hr
8 15
1
 c 1 1   n 1   c   c  
P0          
 n 0 n !    c !      c    
1
 31 1  6 n 1  6 3 12 
P0        
 n 0 n !  4  3!  4  12  6  
1
 1.5 1.5 2
 
2
3
 1.5   4.75  0.21
1
 1  
 1 2 3 
 
c 1 2
1    
Lq    1   P0
c.c !     c 
4 2
1 6  6
 1   (0.21)
3.3!  4   12 
 0.24
 6
(a) Ls  Lq   0.24   1.74
 4
(b) Lq  0.24

Lq 1 0.24 1
(c) Ws      0.29hour
  6 4
4 A supermarket has two girls attending to sales at the counters. If the service time for
each customer is exponential with mean 4 min and if people arrive in Poisson fashion at
the rate of 10 per hour. What is the probability that a customer has to wait for service?
Solution:
Model identification: Since there are two girls and infinite capacity.
Hence this problem comes under the model (M/M/c);(  /FCFS).
Given data:
Arrival rate   10 per hr
1 60
Service rate   per min (i.e)    15 per hr.
4 4
 Number of servers c  2
1
 c 1 1   n 1   c   c  
w.k.t P0          
 n 0 n !    c !      c    
1
 1 1  10 n 2 15  10  
2

P0        
 n 0 n !  15  2!((2 15)  10)  15  
1
 1 1 30 2
   0.6667    0.6667  
n

 n 0 n ! 40 
1
 1 1 
   0.6667   0.3333
n

 n 0 n ! 
P0  (1  0.6667)  0.3333  21  0.5
1

i) The probability that a customer has to wait for the service is

c
 10 
2

   
 15   
P( N s  2)  (0.5) P( N s  c)  P0
 10    
2!1   c !1  
 15  2   c 

 0.6667 
2
1
P( N s  2)  (0.5)   0.1667
1.3333 6
5 A Telephone exchange has two long distance operators. The telephone company finds that
during the peak load, long distance calls arrive in a Poisson fashion at an average rate of
15 per hour. The length of service on these calls is approximately exponentially distributed
with mean length 5 minutes.
(1) What is the probability that a subscriber will have to wait for his long distance call
during the peak hours of the day?
(2) If the subscribers will wait and are serviced in turn, what is the expected waiting time?
Solution:
Model identification: Since there are two operators, infinite capacity.
Hence this problem comes under the model (M/M/c);(  /FCFS).
Given data:
Arrival rate   15 / hr
1
Service rate   per min
5
60
(i.e)    12 / hr
5
Number of servers c  2
 1
 c 1 1   n 1   c   c  
w.k.t P0          
 n 0 n !    c !      c    
1
 1 1  15 n 2 12  15  
2

P0        
 n 0 n !  12  2!((2 12)  15)  12  
1
 1 1 24 2
  1.25  1.25 
n

 n 0 n ! 18 
1
 1 1 
  1.25  2.078
n

 n 0 n ! 
P0  (1  1.25)  2.078  4.3281  0.2311
1

c 1 2
1    
Lq    1   P0
c.c !     c 
2

1.25 1   (0.2311)

1 3 15

(2)2!  24 
2

1.25 1   (0.2311)

1 3 15

4  24 
 1.953  7.111 0.2311  0.8023
 15
Ls  Lq   0.8023   2.0523
 12
i) the probability that a subscriber will have to wait for his service is
c
 15 
c

  
P( N s  2)    P( N s  c)   
12
(0.2311) P0
 15 
2!1  
  
c !1  
 24   c 

1.25
2

P( N s  2)  (0.2311)  0.4814
0.75
ii) If the subscriber will wait and are serviced in turn then the expected waiting time
Ls 2.0523
Ws    0.1368
 15
6 There are three typists in an office. Each typist can type an average of 6 letters per hour. If
letters arrive for being typed at the rate of 15 letters per hour,
(a) What fraction of the time all the typists will be busy?
(b) What is the average number of letters waiting to be typed?
(c) What is the average time a letter has to spend for waiting and for being typed.
(d) what is the probability that a letter will take longer than 20 min waiting to be typed
and being typed? Assume that arrival and service rates follow poisson distribution.
Solution:
This is an (M / M / c) : ( / FIFO) model.
  2.5
  15 / hr &   6 / hr, c  3,   2.5 &   0.833
 c 3
1
 c 1 1   n 1   c   c  
P0          
 n 0 n !    c !      c    
1
 31 1  15 n 6*3  15  
3

       
 n 0 n !  6  3!((6  3)  15)  6  
1
 1  15 0 1  15 1 1  15  2 
           15.625  1  2.5  3.125  15.625
1

 0!  6  1!  6  2!  6  
  22.25  0.0449
1

(a) All the typists will be busy if there are at least 3 customers (letters) in the system
p(n  3)  p(3)  p(4)  p(5)  .....  1  p0  p1  p 2 

 2
p0  2.5 p0
1
p1  p0  2.5 p0 , p2 
2

1!  2!  2
2

P(n≥3) = 1 – [1 + 2.5 + (2.52/2)].(.0449)= 1-0.2974625= 0.7025375= 0.7025

(b) Waiting to be typed (queue)
c 1 2
1    1
Lq  Lq  1   P0 
   2.54 (0.0449)  3.5078
c.c!   
 c  3  6
 15
Ls  Lq   3.5078   6.0078
 6
Ls 6.0078
(c) Ws    0.4005 hr.
 15
   c    t  c 1  
  
   1  e  
 P0 
      
d ) P(Ws  t )  e  t 1  
    
c !1   c  1    
   c  
 
  
1   2.53 1  e( 2)( 0.5)  (0.0449) 
e
6
 
3
1    0.4616
 2.5 


6 1    0.5 

 3 

7 A petrol pump has two pumps. The service times follow the exponential distribution with
mean 4 minutes and cars arrive for service in a Poisson process at the rate of 10 cars per
hour. Find the probability that a customer has to wait for service and what is the
probability that the pumps remain idle?
Solution:
Model identification: Since the petrol bunk has 2 pumps, infinite capacity.
Hence this problem comes under the model (M/M/c);(  /FCFS).
Given data:
Arrival rate   10 per hr
1 60
Service rate   per min (i.e)    15 per hr.
4 4
Number of servers c  2
1
 c 1 1   n 1   c   c  
w.k.t P0          
 n 0 n !    c !      c    
1
 1 1  10 n 2 15  10  
2

P0        
 n 0 n !  15  2!((2 15)  10)  15  
1
 1 1 30 2
   0.6667    0.6667  
n

 n 0 n ! 40 
1
 1 1 
   0.6667   0.3333
n

 n 0 n ! 
P0  (1  0.6667)  0.3333  21  0.5
1

i) The probability that a customer has to wait for the service is

c
 10 
c

   
 15  
P( N s  2)  (0.5) P( N s  c)    P0
 10    
2!1   c !1  
 15  2   c 

 0.6667 
2
1
P( N s  2)  (0.5)   0.1667
1.3333 6
 10 1
ii) Probability of time that a pump is busy     
c 2  15 3
  Probability of time when a pump is idle = 1- Probability of time that a pump is busy
1 2
= 1 
3 3
8. A bank has two tellers working on savings accounts. The first teller handles withdrawals
only while the second teller handles deposits only. It has been found that the service time
distribution for the deposits and withdrawals both is exponential with mean service time
3 minutes per customer. Depositors are found to arrive in a Poisson fashion throughout
the day with mean arrival rate 16 per hour. Withdrawer's also arrive in a Poisson fashion
with mean arrival rate of 14 per hour.
i) What would be the effect on the average waiting time for depositors and withdrawers if
each
teller could handle both withdrawals and deposits?
ii) What would be the effect if this could only be accomplished by increasing the service
time to 3.5 minutes?
Solution:
1
Case(i) :Given  3 min    20 / hr


Average waiting time in

Depositors Withdrawers
the queue
When there is a separate
channel then for the 1 16 2 14
Wq   Wq  
   1  2020  16     2  20  20  14 
(i)depositors 1  16 / hr
1 7
(ii)withdrawers  hr (or ) 12 min  hr (or ) 7 min
5 60
2  14 / hr

If both tellers do the

service then c=2, 
c

  20 / hr   .P0 1.52 . 1  0.45 h or 3.86 min

1 1  1 1
Wq  . .  . .
  1 2  30 / hr  c*c!   
2
20 2.2 1  .75 2 7 7

 c 
1
 
Where
1
 c 1 1   n 1   c   c  
P0          
 n 0 n !    c !      c    
1

 1  1.5 
1.5 
2
1
 
 2  0.25  7

Hence if both tellers do both types of service, the customers get benefited as their waiting
time is considerably reduced.
Case (ii): If both tellers do both types of service with increased service time then
60 120
s=2,    / hr ,   1 2  30 / hr
3.5 7
1
 

P0  1  1.75 
1.75   1
2

1  15
 2x 
 8 
c

   .P0 1 1.75 
2
. 
1 1 7 1 2.86
Wq  .  . . .   0.1907
 c *c!   
2
120 2.2  7
2
15 15
   1  
c 
1
  8

So, if this arrangement is adopted, withdrawers stand to lose as their waiting time is
increased considerably and depositors get slightly benefited.
MODEL - III
The governing equations for the (M/M/1): (GD/ N/ queuing model, the expression for
the steady state probabilities and the average number of customers in the system:
Let N=k (M/M/1): (k /FIFO)
In general, steady state probabilities
01 n1 1
Pn  P with P0 
12 n 0 
01 n1
1 
n 1 1 2 n
Here, n   , n  1, 2,3,
 if n  0,1, 2 k 1
n  
0 if n  k

1
 1 1 1 
when  1, P0    
 
  n 1 
n

n

k 1
1  0 1
k k
1      1  
n 1 1  2 n n 1    n 0    

1
  1
when  1, P0  Lt 
 
1 
k 1
K 1
 1  

 
 1 
 k 1
if   
 
P0   1   
 
 1
 if   
 K 1
    
 n 1 
     
    k 1 
if   
Pn     1     
   
   

 1
if   
 K  1

Average number of customers in the system

If   
      (k  1)   
k 1

 1  
    
n
k k
   
Ls  E ( N )   nPn   n    k 1 
 
n 0            
k 1
n 0
 1     1  1    
  
       
 
If   
k k
1 1 1  k (k  1)  k
Ls  E ( N )   nPn   n  0 1 2  k   
n 0 n 0 k 1 k 1 k 1  2  2

PROBLEMS
1 A small mail –order business has one telephone line and a facility for call waiting for two
additional customers. Orders arrive at the rate of one per minute and each order requires
2 minutes and 30 seconds to take down the particulars. What is the expected number of
calls waiting in the queue? What is the mean waiting time in the queue?
Solution:
Model identification:
since there is only one telephone line and the capacity of the system is finite. Hence this
problem comes under the model (M/M/1);(k /FCFS).
Given Data:
Arrival rate   1/ min
1 5 2  1 5
Service rate     per min , k=3      2.5
 2 5  2/5 2

The expected number of calls waiting in the queue = Lq  Ls 

 (k  1)  k 1
Where Ls   , if    and     1  P0 
1  1   k 1
2.5 4(2.5)4
Ls    0.3333  4.105  4.4383
1  2.5 1  (2.5)4
 (1   )
P0  , if   
1   k 1
1  2.5 1.5
P0    0.0394
1  (2.5) 4
38.06
2
  1  0.0394  0.3842
5
0.3842
The expected number of calls waiting in the queue = Lq  4.4383   3.4778
0.4
Ls 1 4.4383
The mean waiting time in the queue= Wq     2.5  11.55  2.5  9.052
   0.3842
2 Trains arrive at the yard every 15 minutes and the service time is 33 minutes. If the line
capacity of the yard is limited to 5 trains, find the probability that the yard is empty and
the average number of trains in the system, given that the inter arrival time and service
time are following exponential distribution.
Solution:
Model identification:
since there is only one server and the maximum number of customers allowed is 2,
capacity of the system which is finite. Hence this problem comes under the model
(M/M/1);(k/FCFS).
Given Data:
1 1
Arrival rate   per min Service rate   per min
15 33
 33
Capacity of the system k  5    2.2
 15

1
 1  1  2.2
i) Probability that the yard is empty  P0     0.01068

k 1
1  k 1
1  2.26
1  

ii) Average number of trains in the system:
 (k  1)  k 1
Ls   since   
1  1   k 1
k 1

(k  1)  
(or) Ls 

 
  
k 1

1  

 2.2 (5  1)(2.2)6  680.28 
Ls    (1.8333)     4.221
1  2.2 1  (2.2) 6
 112.37 
3 Patients arrive at a clinic according to Poisson distribution at the rate of 30 patients per
hour. The waiting room does not accommodate more than 14 patients. Examination time
per patient is exponential with a mean rate of 20 per hour.
(a) Find the effective arrival rate at the clinic.
(b)What is the probability that an arriving patient will not wait?
(c)What is the expected waiting time until a patient is discharged from the clinic?
Arrival rate   30 / hr., Service rate  20 / hr. , M / M /1: K / FIFO Model.

1- λ 1- 3
μ
   , p0  = 2 =0.00076
 
16
k+1
 
1- λ
μ
1- 3
2

(a) The effective arrival rate is     1  p0   20 1  0.00076   19.98 / hr.

(b) P  a patient will not wait   p  n  0  p0  0.00076

K 1
 3
16

 K  1   16  
(c) Ls 

   (3)    16  13patients nearly
2
  
K 1
3
1   1  
 2
Ls 13
Ws    0.65hr or 39min.
  19.98
4 Consider a single server queueing system with Poisson input, exponential service times.
Suppose the mean arrival rate is 3 calling units per hour, the expected service time is 0.25
hours and the maximum permissible number calling units in the system is two. Find the
steady state probability distribution of the number of calling units in the system and the
expected number of calling units in the system.
Solution:
Model identification:
since there is only one server and the maximum number of calling source is 2, capacity of
the system is finite. Hence this problem comes under the model (M/M/1);(k/FCFS).
Given Data:
Arrival rate   3 / hr
1 1
Service rate    per hr (i.e)   4 / hr
0.25 1/ 4
Capacity of the system k  2
  3
   0.75
 4
i) Steady state Probability distribution of the number of calling units is
Pn , n  0 (i.e) To find P0 , P1 , P2

(1   )  n (1  0.75)(0.75)n
Pn    (0.4324)(0.75)n
1   k 1 1  0.753
P0  (0.4324)(0.75)0  0.4324

P1  (0.4324)(0.75)1  0.3243

P2  (0.4324)(0.75)2  0.2432
ii) The average number of calling units in the system is
k 1

(k  1)  
 (k  1)  k 1  
Ls   since    (or) Ls  
1  1   k 1   
k 1

1  

0.75 (2  1)(0.75)3
Ls    0.81
1  0.75 1  (0.75)3
5 The local one person barber shop can accommodate maximum 5 people at a time
(4 waiting and 1 getting hair cut).Customers arrive according to a Poisson distribution
with mean 5 per hour .The barber cuts hair at an average rate of 4 per hour (exponential
service time)
(i)What percentage of time is the barber idle.
(ii)What fraction of the potential customers are turned away?
(iii)What is the expected number of customers waiting for a hair cut?
(iv)How much time can a customer expect to spend in the barber shop?
Solution:
  5,   4, k  5
 5
1   1  
(i ) p0  Ppercentage of time is the barber idle)     4   0.0888
    k 1    5 51 
1     1
     4  
   
 
 1      5
 
k   5 1  
   
   
5 4
(ii ) P( N  5)     k 1 
 0.2711
      4  5 
6

 1       1    
     4 
 '
(iii) Lq  Ls  where  '   (1  p0 )  Ls  (1  p0 )

  
k 1
5
6
  k  1    6 


       (1  p )  5   4   (1  0.0888)
  k 1  0 6
 1      5
1  
 
   4
15625
6
 5  4096  (1  0.0888)  2.2 customers.
11529
4096
Ls 3.1317
(iv)Ws  '   0.8592 h  51.5 min .
 3.6448

MODEL – IV : (M/M/c);(k/FCFS)
Multi server with finite capacity model.
1
 c 1 n c nc
 1  1 k    
P       
0  n  0 n!    
c!    n  c   c  
 
n
1 
pn    p0 , if 0  n  c
n!   
k
1
pn  n c   p0 , if c  n  k
c c!   

 
c
 
L   1 -  k - c - (k - c)(1 -  )  k - c P
q    c !(1 -  )2 0

 c 1 
Effective arrival rate     c   (c  n) P 
 n0 n


Ls  Lq 

PROBLEMS
1 A group of engineers has two terminals available to aid their calculations. The average
computing job requires 20 minutes of terminal time and each engineer requires some
computation one in half an hour. Assume that these are distributed according to an
exponential distribution. If there are 6 engineers in the group, find
a) The expected number of engineers waiting to use the terminals in the computing center.
b) the total time lost per day.
Solution:
Since there are 2 terminals, also since there are 6 engineers in the group, the capacity of
the system is finite.
Hence this problem comes under the model (M/M/c);(k/FCFS)
Given data:
1
Arrival rate   =2 per hr
1/ 2
1 60
Service rate   per min (i.e)    3 per hr.
20 20
Number of servers c  2
Capacity = k = 6
Expected number of engineers waiting to use in the computing center = Ls

Ls  Lq 

1
 c 1 n c nc
 1  1 k    
Where P       
0  n  0 n!    
c!    n  c   c  
 
1
 1 1 2 n 1 2 2 6 n  2
     2 
P     
0  n  0 n!  3  2!  3  n  2  3  2  
 
1
 4 
2 1  2   1  1 
2 2 3
1 1 
= 1      1             0.5003
 3 2  3  3  3  3  3  
  
 1
And   
c 3

 
c
 
L   1 -  k - c - (k - c)(1 -  )  k - c P
q    c !(1 -  )2 0

1
2  62 62
2 3  1  1  1  
L      (0.5003)  0.0796
2   3    
1 - - (6 2) 1 -
q 3  3   3  
 1
2!1 -   
 3
 c 1   1 
Effective arrival rate     c   (c  n) P   3  2   (2  n) P 
 n0 n  n0 n
 

   3 2   2P0  P1  where P    P
1  0

  2 
   3  2   2  0.5003   0.5003    1.9976
  3 
 Expected number of engineers waiting to use in the computing center = Ls
 1.9976
Ls  Lq   0.0796   0.7455
 3

2. A two person barber shop has 5 chairs to accommodate waiting customers. Potential
customers, who arrive when all 5 chairs are full, leave without entering the barber shop.
Customers arrive at average rate of 4 / hr and spend average of 12 min. in the barber’s
chair. Compute p0 , p1 , p7 .
This is an (M / M / 2) : (7 / FIFO)
1  4  4
  4 / hr,   / min.  5 / hr, c  2, k  7,   0.8,     0.4
12  5 c 2  5 
1 1
 c 1 1  n c k   
n c
  1 1 n 2 7
  
n2

(a) p0    c          
 n0 n !   c ! n c  c   n0 n !  2 2 ! n  2  c 
n n
 
1
 1
 
 1  (0.8)  (0.8) 2 (0.4)0  (0.4)1  (0.4) 2  (0.4)3  (0.4) 4  (0.4)5 
 2 
  2.3114881

1
 p0   0.4289
2.311488
1

(b) p1    p0  0.3431
1


1!

7
1 1
(c) p7  nc   p0  5  0.8  0.4289   0.0014
7

c c!    2 2!

3. A car service station has 2 bays offering service simultaneously. Because of space
constraints, only 4 cars are accepted for servicing. The arrival pattern is Poisson with 12
cars per day. The service time in both the bays is exponentially distributed with  = 8 cars
per day per bay. Find the average number of cars in the service station, the average number
of cars waiting for service and the average time a car spends in the system.
This is a multiple server model with finite capacity. λ=12/day and μ=8/day, c=2, K=4
1
 
 c 1 n c n c  1
1  1 k    1.5 1 2 
p0          
  1.5  1   0.75    0.75  
 
2
 1 
 n 0 n !    c !    n c      1 2  
 
 
P0  0.1960 ,
c
  
Lq  P0   1   k c   k  c 1     k c  where  
2 
   c !1    c
  0.75 
Lq =0.1960 1.5  2  
 1- 0.75 -2  0.25 0.75   =0.4134car.
2 2 2

 2  0.25  
 
c 1 1
Ls = Lq  c    c  n  pn  0.4134  2    2  n  pn  2.4134   2 p0  p1 
n 0 n 0

1
1  12 
p1    (0.1960) , if 0  n  c  2
1!  8 
 0.294
LS  2.4134   2(0.1960)  0.294  1.73 cars

Ls  c 1 
Ws = where     c    c  n  pn 
  n 0 
 8  2   2 p0  p1  
 8  2  (2(0.1960)  0.294  10.512
1.73
Ws   0.1646 day
10.512
Average time that a car has to spend in the system = 0.16
Impatience Customer
Ls , Lq , Ws & Wq for queues with impatience customer, where the arrival rate is inversely
propotional to the number of customers in the system:

Take n  bn   , n  0,1, 2,...
n 1
And n   , n  1, 2,3,...
1
P0  
01...n 1
1 
n 1 1  2 ... n

1

              
        
1              ...
1 1 2 1 2 3
  
1
 2 3
  
  
1           ...
1! 2! 3!
1 
 where,   1

  

e
 
P0  e  where,   1

Now,
01...n1
Pn  P0 , n  1, 2,3,....
12 ...n
 1 1 1 n
Pn  P0  n 1   ...    e  , n  0,1, 2,3,...
 2 3 n n!


Ls   nPn   
n 1 



Lq   (n  1) Pn   e   1
n 1 