Non-Markovian Queues and Queue Networks

1
MISRIMAL NAVAJEE MUNOTH JAIN ENGINEERING COLLEGE
DEPARTMENT OF MATHEMATICS
PROBABILITY AND QUEUING THEORY (MA 2262)
SEMESTER-IV
QUESTION BANK V
UNIT V NON-MARKOVIAN QUEUES AND QUEUE NETWORKS
PART-A
Problem: 1
Write down Pollaczek - Khinchine formula.
Solution:
i) Average number of Customers in the system
( )
2 2 2
2 1
o

+
= +

where

= .
ii) Average queue length
( )
2 2 2
2 1
o

+
=

Problem: 2
Write the steady-state equations (flow balance equations) for a two-station sequential Queue with
blocking.
Solution:
0,0 2 0,1
p p =
1 1,0 2 1,1 0,0
p p p = +
2 0,1 1 1,0 2 ,1
( )
b
p p p + = +
1 2 1,1 0,1
( ) p p + =
2 ,1 1 1,1 b
p p =
Problem: 3
Write the flow-balance equations of open Jackson networks.
Solution:
0,1,2,...
1
k
j j i ij j k
i
r P
=
=
= +

where P
ij
is the probability that a departure from server i joins the Queue at server j.
Problem: 4
Write the flow-balance equations of closed Jackson networks.
Solution:
0,1,2,...
1
k
j i ij j k
i
P
=
=
=

Non-Markovian Queues and Queue Networks

2
where P
ij
is the probability that a departure from server i joins the Queue at server j.
Problem: 5
State the characteristics of Jackson networks.
Solution:
(a) Arrivals from outside through node i follow a Poisson process with mean arrival rate r
i
.
(b) Service times at each channel at node i are independent and exponentially distributed
with parameter
i

(c) The probability that a customer who has completed service at node i will go next to node j
(Routing probability) is P
ij
, i= 1, 2, 3, .k and j = 0, 1, 2, 3, k. and P
i0
denotes the probability
that a customer will leave the system from node i. P
ij
is independent of the state of the system.
If
j
is the total arrival rate of customers to node j,
0,1,2,...
1
k
j j i ij j k
i
r P
=
=
= +

Problem: 6
Distinguish between open Jackson networks and closed Jackson networks
Solution: We have,
0,1,2,...
1
k
j j i ij j k
i
r P
=
=
= +

In all general cases where, r

i
0 for any i or P
i0
0for any i, the Jackson networks are referred to as open
Jackson networks. In the case of closed Jackson networks, r
i
= 0 for all i and P
i0
= 0for all i.
PART-B
Problem: 7
Derive Pollaczek - Khinchine relation for ( ) ( ) / /1 : / M G GD
Solution:
In order to determine the mean queue length L
q
and mean waiting time Wq for this system the
following technique is used.
Let ( ) f t = Probability distribution of service time t with mean ( ) E t and Variance ( ) V t .
n =no. of customers in the system just after a customer departs.
t = service time of the customer following the one already departed.
1
n = The number of customers left behind the next departing customer.
1
0 n k if n = =
( ) 1 0 n k if n = + >
Where k = 0,1,2.is the number of arrivals during the service time.
Alternatively of 1 0 if n o = =
0 0 if n = >
1
1 (1) n n k o = + +
Non-Markovian Queues and Queue Networks
3
Then taking expectation
( ) ( ) ( ) ( )
1
1 E n E n E E k o = + + Since
( ) ( )
1
E n E n = in steady state ( ) ( ) 1 E E k o =
Squaring (1) both sides
( ) ( )
2 2
1
1 n n k o = + + (

( ) ( ) ( )
2
2 2
1 2 1 2 2 1 . n k n k n k o o o = + + + + +
( ) ( )
2
2 2
1 2 1 2 2 2 n k n k n k o o o o = + + + + +
Since o can take values 0 or 1only
2
0 and n o o o = =
( )
2
1 2 2
2 1 2 1 2 2 n n k k n k k o o o = + + + + +
( ) ( )
2 2
2 1 2 1 2 1 n k n k k k o = + + + + +
( ) ( )
2
2 1 2
2 1 2 1 2 1 n k n n k k k o = + + +
Taking expectation on both sides
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )
2 1 2
2 1 (2 ) 1 2 1 E n E k E n E n E k E E k E k o
2
= + + + (

( )
( ) ( ) ( ) ( ) ( )
( ) ( )
2
2 2 1 1
2 1
E k E k E E k
E n
E k
o + +
=

( ) ( ) 1 E E k o =
( )
( ) ( ) ( ) ( ) ( )
( ) ( )
2
2 1 2 1 1
2 1
E k E k E k E k
E n
E k
+ + (

=

. ( )
( ) ( ) ( )
( ) ( )
2 2
2
2 1
E k E k E k
E n
E k
+
=

Now is order to determine ( ) E n , the values of

( ) ( )
2
and E k E k are to be computed. Since the
arrivals follow Poisson distribution
( ) ( ) ( )
0
/ E k E k t f t dt

=
}
Now ( ) ( ) ( ) ( )
2
2
/ , / E k t t E k t t t = = +
( ) ( ) ( )
2 2
/ / / V k t E k t E k t =
( ) ( ) ( )
0
E k t f t dt E t

= =
}
( ) ( ) ( )
2 2
0
/ E k E k t f t dt

=
}
( ) ( )
2
0
t t f t dt

(
= +

}
( ) ( ) ( ) ( ) ( )
2 2 2 2
E t E t V t E t E t
( ( = + = + +

( ) ( ) ( ) ( )
2 2 2
E k V t E t E t ( = + +

( )
( ) ( ) ( ) ( ) ( )
( )
2 2 2 2
2
2 1
V t E t E t E t E t
E n
E t

( + + +

=
(

Non-Markovian Queues and Queue Networks
4
( ) ( ) ( ) ( )
( ) ( )
2 2 2 2 2
2 2
2 1
V t E t E t E t
E t

+ +
=

( ) ( ) ( ) ( )
( ) ( )
2 2 2
2 1
2 1
V t E t E t E t
E t

+ + (

=

( )
( ) ( )
( ) ( )
2 2 2
2 1
E t V t
E t
E t

+
= +

( ) ( )
( ) ( )
( ) ( )
2 2
2 1
s
E t V t
L E n E t
E t

( +

= = +

s
s
L
W =
( ) 1/ where is the rate of servive. E t =
Then
( )
2 2 2
2 2 2
(1/ )
( )
/
2 1 / 2(1 )
s
L
o
o


( +
+

= + = +

where
2
/ and ( ) V t o = =
2 2 2
( )
2(1 )
q s
L L
o

+
= =

Problem: 8
A one man barber shop takes exactly 25 minutes to complete one haircut. If customers arrive at
the barber shop in a Poisson fashion at an average rate of one every 40 minutes, how long on the
average a customer spends in the shop? Also find the average time a customer must wait for
service.
Solution:
Since the service time T is a constant=25 minutes, T follows a probability distribution with
E(T)=25, Var(T)=0.
1 = for every 40 minutes
1
40
= per minutes
By Pollaczek-Khinchine formula,
( )
( ) ( ) ( )
( ) ( )
2
2
2
2
var
2 1
1
0 25
25 40
1 40
2 1 .25
40
1
.625
25 25 625 4 5 25 55
1600
15 40 40 1600 3 8 48 48
2
40
s
T E T
L E T
E T

(
+

= +

| |
( +
|

\ .
= +
| |

|
\ .
= + = + = + =
| |
|
\ .
By Little formula,
Non-Markovian Queues and Queue Networks
5
55
55
48
40
1
48
40
s
S
L
W

= = =
45.8 = Minutes
( )
( )
1
1
45.8
45.8 25
q s
W W
E T
E T

=
(
= =
(

=
20.8 = Minutes
Hence, a customer has to spend 45.8 minutes in the shop and has to wait for service for 20.8
minutes on the average.
Problem: 9 Suppose a one person tailor shop is in business of making mens suits. Each suit
requires four district tasks to be performed before it is completed. Assume all four tasks must be
completed on each suit before another is started. The time to perform each task has an
exponential distribution with a mean of 2 hr. If orders for a suit come at the average rate 5.5 per
week (assume an 8hr day, 6 day week), how long can a customer expect to wait to have a suit
made?
Solution: Given: 5.5 = per week
5.5
6 8
=

orders/hr
=0.1149 orders/hr
The server time for each task
1
k
= where 4 k =
i.e.,
1
2 .
4
hr

= = (given)
1
8
hr =
=0.125 orders / hr.
0.1149
0.9192
0.125

= = =
Since the service time T is constant then
2
0 o =
Here the average service rate is the average service rate to complete a suit.
The expected waiting time of a customer (in the system) is
Non-Markovian Queues and Queue Networks
6
( )
2 2 2
1
2 1
s
W
o

+
= +

Problem: 10 In a heavy machine shop, the overhead crane is 75% utilized. Time study observations
gave the average slinging time as 10.5 minutes with a standard deviation of 8.8 minutes. What is the
average calling rate for the services of the crane and what is the average delay in getting service? If the
average service time is cut to 8.0 minutes, with a standard deviation of 6.0 minutes, how much
reduction will occur, on average, in the delay of getting served?
Solution: This is a ( ) ( ) / /1 : / M G FIFO Process
Given: Utilization rate
75 3
75%
100 4
= = =
i.e.,
3
4
=
Average service time E(T)=10.5 min
( )
1 1
10.5 E T
= =
We know that

=
=
3 1
0.0714min
4 10.5

| || |
= =
| |
\ .\ .
i.e., average calling rate for the services of the crane
0.0714min 4.286 = = hour
To find the average delay in getting service
Here
3
8.8, 0.0714, 0.75
4
o = = = =
Average delay in getting service
( )
( ) ( ) ( )
( )( )
2 2 2
2 2 2
0.0714 8.8 0.75
26.815 min
2 1 2 0.0714 1 0.75
q
W
o

+ +
= = =

i.e., The average service time is cut to 8.3 minutes then
( )
( )
1 1
min
8.0
0.0714min
0.0714
0.5712
1/ 8
6min
E T

o
= =
=
= = =
=
Non-Markovian Queues and Queue Networks
7
Average delay in getting service
( )
( ) ( ) ( )
( )( )
2 2 2
2 2 2
0.0714 6 0.5712
2 1 2 0.0714 1 0.5712
8.326 min
q
W
o

+ +
= =

=
The Average waiting time has a reduction of
26.8-8.32=18.5 minutes
( )
( )( )
2
0 0.9192 1
53.42
2 0.1149 0.0808 0.125
+
= + = hrs =1.113 Weeks .
Problem: 11 A car manufacturing plant uses one big crane for loading cars into a truck. Cars arrive for
loading by the crane according to a Poisson distribution with a mean of 5 cars per hour. Given that the
service time for all cars is constant and equal to 6 minutes determine , ,
S q S q
L L W and W .
Solution: The given problem is in ( ) ( ) / /1 : / M G FIFO model.
5cars/hr = cars/hr.
The service time T is constant with mean | |
1
6 E T

= = min
Since the service time T is constant then var[T]=
2
0 o =
1 1 1
6 min 6 .
60 10
10
5 1
10 2
hr hr
hr
hr

= = =
=
= = =
Using k formula, we have
( )
( )
( )
2 2 2
2 1
0 1/ 4
1 3
0.75 /
2 2 1 1/ 2 4
3 1
4 2
0.25 /
s
q s
L
car hr
L L
car hr
o

+
= +

+
= + = =

=
=
=
( )
( )
3/ 4 3
9
1/ 5 20
s
s
L
W hr

= = = = minutes
Non-Markovian Queues and Queue Networks
8
( )
( )
1/ 4 1
3
1/ 5 20
q
q
L
W hr

= = = = minutes.
Problem: 12 A car wish facility operates with only one bay. Cars arrive according to a Poisson
distribution with a mean of 4 cars per hour and may wait in the facilitys parking lot if the bay is
busy. The parking lot is large enough to accommodate any number of cars. Find the average
number of cars waiting in the packing lot, if the time for washing and cleaning a car follows.
(a) Uniform distribution between 8 and 12 minutes
(b) A normal distribution with mean 12 minutes and S.D.3minutes.
(c) A discrete distribution with values equal to 4,8 and 15 minutes and corresponding
probabilities 0.2,0.6 and 0.2.
Solution:
Mean = 4 = / hour =
4
60
per minute
1
15
= per minute
Let T be a continuous random variable.
Then, E(T) = mean of the uniform distribution
( ) ( ) ( )
1 1 1
8 12 20 10
2 2 2
a b = + = + = =
Var (T)
( )
2 1
12
b a =
( ) ( ) ( )
2 2 1 1 1 4
12 8 4 16
12 12 12 3
= = = =
By the Pollaczek-Kninchine formula,
( )
( ) ( ) ( )
( ) ( )
( )
2
2
2
2
var
.
2 1
1 4
10
1 15 3
.10
1 15
2 1 .10
15
1 4
100
2 225 3
2 3
2 1
3
1 304
.
2 2 1 304 3 2 152 302
225 3
. .
1 3 3 225 3 2 3 225 225
2
3
s
T E T
L E T
E T

(
+

= +

| | (
+
|
(
\ .
= +
| |

|
\ .
(
+
(

= +
| |

|
\ .
= + = + = + =
| |
|
\ .
Non-Markovian Queues and Queue Networks
9
1.342 = Cars
~ Car
By Littles formula,
1
15
1.342
1
10
q s
L L

=
=
0.675 = cars
~ 1 car
(b)
1
,
15
= ( ) 12 E T = min, and Var(T)=9
( )
1 1
12 E T
= =
By the Pollaczek-Kninchine formula,
( )
( ) ( ) ( )
( ) ( )
2
2
2
2
var
.
2 1
1
1
9 12
.153
1 4 4 153 15 15
225
.12 .
1 3 15 5 5 225 6
2 1 .12 2
15 15
s
T E T
L E T
E T

(
+

= +

| |
( +
|

\ .
= + = + = +
| | | |

| |
\ . \ .
4 153
2.5
5 90
= + = Cars.
By Littles formula,
1
12
15
2.5 2.5 2.5 0.8 1.7 2
1
15
12
q s
L L
Cars Cars

=
= = = = ~
(c) The service time T follows the discrete distribution given below.
T : 4 8 15
P(T) : 0.2 0.6 0.2
( ) ( )
( ) 4 0.2 8 0.6 15 0.2
0.8 4.8 3
TP T E T =
= + +
= + +

8.6 = minutes
Non-Markovian Queues and Queue Networks
10
( ) ( )
( ) ( )
2
2
4 0.2 8 0.6 15 0.2 0.8 4.8 3 8.6minutes
T P T
2
E T =
= + + = + + =

( ) ( ) ( )
( )
2
2
2
var
86.6 8.6 12.64
E T E T T = (

= =
BY Pollaczek-Khinchine formula,
( )
( ) ( ) ( )
( ) ( )
( )
2
2
2
var
.
2 1
1
12.64 8.6
1
225
8.6
1 15
2 1 8.6
15
s
T E T
L E T
E T

(
+

= +

(
+

= +
(

(

| |
1
12.64 73.96
86.6 15
225
0.573 0.573 1.024 1
8.6 225 12.8
2 1
15
Car
+
= + = + = ~
(

(

By Little formula,
q s
L L

=
( )
1
1
5
1.024
1
86
E T

(
= =
(

8.6
1.025 0.45
5
Car = =
Problem: 13 An automatic car wash facility operates with only one bay; Cars arrive according to
a Poisson process with mean of 4 cars per hour and may wait in the facilitys parking lot if the
bay is busy. If the service times for all cars is constant and equal to 10 minutes.
Determine , , ,
q q s q
L L W W .
Solution:
4 = per hour
4
60
= per minute
1
15
= per minute
Given service time T is a constant.
T follows a distribution with E(T)=10 and Var(T)=0.
Non-Markovian Queues and Queue Networks
11
By Pollaczek - Khinchine formula,
( )
( ) ( ) ( )
( ) ( )
2
2
2
2
var
.
2 1
1
0 10
1 15
10
1 15
2 1 .10
15
100
10 10 100 15 20 4
225
15 10 15 15 225 10 15 3
2
15
s
T E T
L E T
E T

(
+

= +

| |
( +
|

\ .
= +
| |

|
\ .
= + = + = =
| |
|
\ .
By Little formula,
4
15
3
s
S
L
W

= = 20 = minutes
( )
( )
1
1
20
20 10
q s
W W
E T
E T

=
(
= =
(

=
10 = minutes.
i.e., a customer has to spend 20 minutes in the system and 10 minutes in the queue.
.
1 10
10
15 15
q q
L W =
= =
Problem: 14 In a big factory, there are a large number of operating machines and two
sequential repair shops, which do the service of the damaged machines exponentially with
respective rates of 1/hour and 2/hour. If the cumulative failure rate of all the machines in the
factory is 0.5 / hour, find (i) the probability that both repair shops are idle, (ii) the average
number of machines in the service section of the factory and (ii) The average repair time of a
machine.
Solution: Given 0.5 = / hour
1
2
= per hour
1
= 1 per hour
2
= 2 per hour
Non-Markovian Queues and Queue Networks
12
The situation in this problem is comparable with 2-stage Tandem queue with single server at
each state.
(i) P(both the service stations are idle)
( )
0 0
1 1 2 2
0 0
0, 0
. 1 . . 1
1/ 2 1/ 2 1/ 2 1/ 2
1 1
1 1 1 1
1 1 1 3 3
1
2 4 2 4 8
P


=
| | | | | | | |
=
| | | |
\ . \ . \ . \ .
| | | || | | |
=
| | | |
\ . \ .\ . \ .
| || | | |
= = =
| | |
\ .\ . \ .
(ii) The average number of machines in service
1 2
1
1/ 2 1/ 2 1/ 2 1 4
2
1
1 1
1/ 2 3/ 2 3 3
1 2
2 2


= +

= + = + = + =

(iii) The average repair time
1 2
1 1
1 1 1 1 8
1 1 1 3 3
1 2
2 2 2 2

= +

= + = + =
| | | |

| |
\ . \ .
Problem: 15 A TVS company in Chennai containing a repair section shared by a large number
of machines has 2 sequential stations with respective service rates of 3 per hour and 4 per hour.
The cumulative failure rate of all the machines is 1 per hour. Assuming that the system behavior
can be approximated by the above 2-stage tendon queue, find
(i) the probability that booth the service stations are idle (free)
(ii) the average repair time including the waiting time.
(iii) the bottleneck of the repair facility.
Solution:
STEP-1: Model Identification
The current situation comes under the sequence queue model, Since any number of machines can
be repaired, each station comes under the model ( ) ( ) / /1 : / M M FCFS
STEP 2: Given Data
Cumulative failure rate 1 =
Service rate of station I
1
3 =
Non-Markovian Queues and Queue Networks
13
Service rate of station II
2
4 =
STEP 3: To find the following
(1) The probability that both the servive stations are idle.
(2) The average repair time and waiting time
(3) The bottleneck of the repair facility.
STEP 4: Required Computations
(i) P( m customers in the I station and n customers in the II station
1 1 2 2
1 1 2 2
0 0
00
1 1
1 1 1 1 2 3 1
1 1
3 3 4 4 3 4 2
m n
mn
P
P


| | | || | | |
=
| | | |
\ . \ .\ . \ .
| | | || | | |
= = =
| | | |
\ . \ .\ . \ .
(ii) The average number of machines in service at the system(both the stations)
1 2
1 1 1 1 5
3 1 4 1 2 3 6
s
L


= +

= + = + =

= 1 machine (approximately)
Average repair time (including waiting time)
1 2
1 1

= +

5
6
= hours =50 minutes.
(iii) Since
1 2
1
3


= >
1
4
=
We get the service station 1 is the bottleneck of the repair facility.
Problem: 16 In the Airport reservation section of a city junction, there is enough space for the
customers to assemble, form a queue and fill up the reservation forms. There are 5 reservation
counters in front of which also there is enough space for the customers to wait. Customers arrive
at the reservation counter section at the rate of 40 per hour and takes one minute on the average
to fill up the forms. Each reservation clerk takes 5 minutes on the average to complete the
business of a customer in an exponential manner.(i) Find the probability that a customer has to
wait to get the service in the reservation counter section (ii) Find the total waiting time for a
customer in the entire reservation section. Assume that only those who have the filled up
reservation forms will be allowed into the counter section.
Solution:
Non-Markovian Queues and Queue Networks
14
The queuing system in the form-filling portion is only a M/M/1 model, since each customers is
served by himself or herself (viz., one server)
For this system, 40 = / hour and 60 = /hour.
( )
40 40
2
60 40 20
s
E N


= = = =

( )
1 1
20
s
E W

= =

hour or 3 min
The queuing system in the reservation counter section is an M/M/s model with 40 = /hour,
12 = /hour and 5 s = [Since the output of the M/M/1 system in the same as the input of that
system, by Burkes theorem and the output of this system (namely, 40/hour) becomes the input
of the M/M/s system]
(i) P(a customer has to wait in the counter section)
( )
2
0
.
, .... 1
1
P
s
s

| |
|
\ .
=
| |

|
\ .
Where
( )
( )
( )
1
1
0
0
1
5
4 1
0
10 1 2 3 4
/ 1
.
1
10
1 10 3
.
3 5! 1 2/ 3
100000
1 10 1 10 1 10 1 10 1 10
243
1 0! 3 1 3 2 3 3 3 4 3
120
r s
s
r
r
r
P
r
s
s
r

=
(
(
| |
(
= +
`
|
(
| |
\ .

(
|

\ . )
(
| |
(
|
| |
\ .
(
+
|
(
\ .
(

| | | | | | | | | |
= + + + + +
| | | | |
\ . \ . \ . \ . \ .

1
3

(
(
(
| |
(
|
(
\ .
1
10 100 1000 1 100000 100000 3
1
3 18 162 24 81 243 120

( | || |
= + + + + + +
| | (
\ .\ .
| |
| |
1
1
1
10 100 1000 1 10000
1 10.288
3 18 162 24 81
21.206 10.288
31.494 0.032

(
= + + + + +
(

= +
= =
Non-Markovian Queues and Queue Networks
15
( ) 1 Required probability
( )
5
10
0.032
3
2
5! 1
3
| |
|
\ .
=
| |

|
\ .
( )
( )
| |
( )
1
0 2
100000
0.032
243
0.3292
1
120
3
/ 1
!
1
s
s
E N P
ss
s

+
| |
|
\ .
= =
| |
|
\ .
= +
| |

|
\ .
( )( )
( )
( )( )
( )
| |
6
10
1 10 3
0.032
2 5 5! 3
1
3
1000000
1 10 729
0.032
1 5 120 3
3
1
131.69 3.3333 0.2195 3.3333 3.5528
600
| |
|
\ .
= +
| |

|
\ .
( | |
| (
\ .
( = +
| |
(
|
(
\ .
= + = + =
( ) | | ( )
1 1
3.5528
40
0.0888
s s
E W E N
hour

= =
=
5.329 = minutes
Total waiting time of a customer in the entire reservation room =6+5.329=11.3292 minutes.
Problem: 17 The last two things that are done to a car before its manufacture is complete are
installing the engine and putting on the tires. An average of 54 cars per hour arrives, requiring
these two tasks. One worker is available to install the engine and can service an average of 60
cars per hour. After the engine is installed, the car goes to the tire station and waits for its tires to
be attached. Three workers serve at the tire station. Each works on one car at a time and can put
tires on a car in an average of 3 minutes. Both inter arrival times and service times are
exponential.
(i) Determine the mean queue length at each work station.
(ii) Determine the total expected time a car spends waiting for service.
Solution:
1 1 2 2
54, 1, 60, 3, 20 s s = = = = =
Non-Markovian Queues and Queue Networks
16
Since
1 2
3 and < < , neither queue will below up and Jacksons theorem is applicable.
For stage 1 (engine)
( )
( )
1 1
2
2
54
0.90
60
0.90 0.81
8.1
1 1 0.90 0.1
1 1
8.1 0.15
54
q
q q
s
L cars
W L hour

= = =
= = = =

= = =
For stage 2 (tires)
( )
( )
( )( )
( )
2 2
54 54
0.90
3 20 60
3 0.83
0.83 0.90
7.47
1 0.90
1 1
7.47 0.138
54
q
q q
s
j
L cars
W L hour

= = = =
> =
= =

= = =
Thus, the total expected time a car spends waiting for engine installation an tires is
0.15+0.138=0.288 hour.
Problem: 18 For a 2- stage (service point) sequential queue model with blockage, compute the
average number of customers in system and the average time that a customer has to spend in the
system, if 1 = ,
1 2
2 1 and = =
Solution:
Given
1 2
1, 2, 1 = = = .
The balanced equations are
State Rate that the process leaves
(0,0)
00 2 01
P P = (1)
(1,0)
1 00 00 2 11
P P P = + (2)
(0,1) ( )
2 01 1 10 1 1 b
P P P + = + (3)
(1,1) ( )
1 2 11 01
P P + = (4)
(b,1)
2 1 1 11 b
P P = (5)
00 10 01 11 1
1
b
P P P P P + + + + = (6)
(1)
00 01
P P = (7)
(2)
10 00 11
2P P P = + (8)
(3)
01 10 1
2 2
b
P P P = + (9)
(4)
11 01
3P P = (10)
(5)
1 11
2 Pb P = (11)
Non-Markovian Queues and Queue Networks
17
(7) & (10)
00 01 11
3 P P P = = (12)
(12) & (1)
1
00 01 11
3
3
2
b
P P P P = = = (13)
(8)
10 00 00
1
2
3
P P P = +
10 00
4
2
3
P P =
10 00
2
3
P P = (14)
(6)
00 10 01 11 1
1
b
P P P P P + + + + =
00 00 00 00 00
2 1 2
1
3 3 3
P P P P P + + + + = by (13) & (14)
00
2 1 2
1 1 1
3 3 3
P
(
+ + + + =
(

00
00
5
2 1
3
11
1
3
P
P
(
+ =
(

(
=
(

(13)
01
3
11
P =
11 00
1 1 3 1
3 3 11 11
P P
| |
= = =
|
\ .
i.e.,
(14)
10
2 3 2
. .,
3 11 11
P i e
| |
= =
|
\ .
(13)
1 00
2
3
b
P P =
2 3
3 11
| |
=
|
\ .
1
2
11
b
P =
( )
01 10 11 1
2
3 2 1 2
2
11 11 11 11
5 3 5 6
2 1
11 11 11 11
b
L P P P P = + + +
| |
= + + +
|
\ .
| |
= + = + =
|
\ .
00
3
11
P =
11
1
11
P =
10
2
11
P =
Non-Markovian Queues and Queue Networks
18
( )
( ) ( )
( )
01 10 11 1
00 01 00 01
2
1 1 11
3 3 6/11 6
1
11 11
b
P P P P
L
W
P P P P
+ + +
= =
+ +
= = =
| |
+
|
\ .
Problem: 19 For a 2-stage (service point) sequential queue model with blockage, compute
s
L
and
1 2
, 1, 1 2
S
W if and = = = .
Solution: Given:
1 2
1, 1, 2 = = =
The balanced equations are
( ) ( )
( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
( )
( ) ( )
( )
( )
00 2 01
1 10 00 2 11
2 01 1 10 2 1
1 2 11 01
2 1 1 11
00 10 01 11 1
00 01
10 00 11
01 10 1
11
0, 0 ..... 1
1, 0 ..... 2
0,1 ..... 3
1,1 ..... 4
,1 ..... 5
1 ..... 6
1 2 ..... 7
2 ..... 8
3 2 ..... 9
3
b
b
b
P P
P P P
P P P
P P
b P P
P P P P P
P P
P P P
P P Pb
P P





=
= +
+ = +
+ =
=
+ + + + =
=
= +
= +
= ( )
( )
( ) ( ) ( )
01
1 11
00 01 11
..... 10
2 ..... 11
7 & 10 2 6 ..... 12
b
P P
P P P
=
= =
( ) ( ) ( )
( )
( )
00 01 11 1
10 00 00
10 00
12 & 11 2 6 12 ..... 13
1
8
3
4
..... 14
3
b
P P P P
P P P
P P
= = =
= +
=
( ) ( ) ( )
00 00 00 11 00
00
00
4 1 1 1
6 1 13 & 14
3 2 6 12
4 1 1 1
1 1
3 2 6 12
37
1
12
P P P P P by
P
P
+ + + + =
(
+ + + + =
(

(
=
(

00
12
37
P =
Non-Markovian Queues and Queue Networks
19
( )
01 00
1
13
2
1 12
2 37
P P =
| |
=
|
\ .
( )
11 00
1
13
6
1 12
6 37
P P =
| |
=
|
\ .
( )
1 00
1
13
12
1 12
12 37
b
P P =
| |
=
|
\ .
( )
10 00
4
14
3
4 12
3 37
P P =
| |
=
|
\ .
( )
01 10 11 1
2
6 6 2 1 22 3 28
2 2
37 37 37 37 37 37 37
b
L P P P P = + + +
| | | |
= + + + = + =
| |
\ . \ .
( )
( ) ( )
( )
00 01
28/ 37 28/ 37 14
12 6 18/ 37 9
1
37 37
L
W
P P
= = = =
+ | |
+
|
\ .
Problem: 20 There are two salesmen in a shop, one in charge of receiving payment and the other
in charge of delivering the items. Due to limited availability of space, only one customer is
allowed to enter the shop, that too when the clerk is free. The customer who has finished his job
01
6
37
P =
11
2
37
P =
1
1
37
b
P =
10
16
37
P =
Non-Markovian Queues and Queue Networks
20
has to wait there until the delivery section becomes free. It customers arrive in accordance with a
Poisson process at rate 1 and the service times to two clerks are independent and have
exponential rates of 1 and 3, find
(i) the proportion of customers who enter the ration shop
(ii) the average number of customers in the shop and
(iii) the average amount of time that an entering customer spends in the shop.
Solution: Given :
1 2
1, 1, 3 = = =
The balanced equations are
( ) ( )
( ) ( )
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
( )
( ) ( )
( )
( )
00 2 01
1 10 00 2 11
2 01 1 10 2 1
1 2 11 01
2 1 1 11
00 10 01 11 1
00 01
10 00 11
01 10 1
11
0, 0 ..... 1
1, 0 ..... 2
0,1 ..... 3
1,1 ..... 4
,1 ..... 5
1 ..... 6
1 3 ..... 7
3 ..... 8
4 3 ..... 9
4
b
b
b
P P
P P P
P P P
P P
b P P
P P P P P
P P
P P P
P P Pb
P P





=
= +
+ = +
+ =
=
+ + + + =
=
= +
= +
= ( )
( )
( ) ( ) ( )
01
1 11
00 01 11
..... 10
3 ..... 11
7 & 10 3 12 ..... 12
b
P P
P P P
=
= =
( ) ( ) ( )
( )
( )
00 01 11 1
10 00 00
10 00
12 & 11 3 12 36 ..... 13
1
8
4
5
..... 14
4
b
P P P P
P P P
P P
= = =
= +
=
( ) ( ) ( )
00 00 00 00 00
00
00
00
5 1 1 1
6 1 13 & 14
4 3 12 36
5 1 1 1
1 1
4 3 12 36
97
1
36
36
97
P P P P P by
P
P
P
+ + + + =
(
+ + + + =
(

=
=
Non-Markovian Queues and Queue Networks
21
( )
( )
01 00
11 00
1 00
10 00
1 1 36 12
13
3 3 97 97
1 1 36 3
12 12 97 97
1 1 36 1
36 36 97 97
5 5 36 45
14
4 4 97 97
b
P P
P P
P P
P P
| |
= = =
|
\ .
| |
= = =
|
\ .
| |
= = =
|
\ .
| |
= = =
|
\ .
The proportion of customers entering the shop
00 01
36 12 48
97 97 97
P P = + = + =
(ii) ( )
01 10 11 1
2
b
L P P P P = + + +
12 45 3 1 57 8
2
97 97 97 97 97 97
| |
= + + + = +
|
\ .
65
97
=
(iii)
( )
00 01
65/ 67
36 12
1
97 97
L
W
P P
= =
+ | |
+
|
\ .
65 97 65
97 48 48
| || | | |
= =
| | |
\ .\ . \ .
Problem: 21 Consider a system of two servers where customers from outside the system arrive
at server 1 at a Poisson rate 4 and at server 2 at a Poisson rate 5. The service rates 1 and 2 are
respectively 8 and 10. A customer upon completion of service at server 1 is equally likely to go
to server 2 or to leave the system (i.e.,P
11
=0,P
12
=1/2); whereas a departure from server 2 will go
25 percent of the time to server 1 and will depart the system otherwise (i.e.,P
21
=1/4,P
22
=0).
Determine the limiting probabilities, L and W.
Solution: The total arrival rates to servers 1 and 2 call them
1 2
and - Can be obtained from equation
1
1
k
j i ij
i
r P
=
= +

That is, we have

2
1 1 1 11 2 21
1
2 11 21
4 4
1 1
4 0 0,
4 4
i i
i
P P P
P P

=
= + = + +
(
= + + = =
(

Non-Markovian Queues and Queue Networks

22
( )
| |
( )
1 2
2
2 2
1
1 12 2 22
1 12 22
2 1
1
4 ...... 1
4
5
5
1
5 0 1/ 2, 0
4
1
5 ...... 2
2
i i
i
P
P P
P P

=
= +
= +
= + +
= + + = =
= +

( )
1 1
1
1 1
1
1 1
1 4 5
4 2
5 1
4
4 8
1 21
8 4
7 21
8 4

(
= + +
(

= + +
=
=
( )( )
( )
1
2
21 8
3 2 6
4 7
1
5 6 5 3 8
2

| || |
= = =
| |
\ .\ .
= + = + =
Given
1 2
1 2
1 2
6 3 8 4
8, 0, ,
8 4 10 5



= = = = = =
Hence P[n at server 1, m at server 2]
( ) | |
1 2
1 1 2 2
1 1 2 2
1 2
1 2
1 1 2 2
1 1
3 3 4 4 3 1 16 1 3
1 1
4 4 5 5 4 4 25 5 125
6 8 6 8
3 4 7
8 6 10 8 2 2
1 1
7 4 5
9
7
9
n n
q
q
L
W L

| | ( ( (
=
| ( ( (
\ .
| |
| | | | | | | || || || |
= = =
|
| | | | | | |
\ . \ . \ . \ .\ .\ .\ .
\ .
= +

= + = + = + =

= = = +
=

Non-Markovian Queues and Queue Networks

23
Problem: 22 In a departmental store there are 2 sections, namely, grocery section and perishable
(vegetables and fruits) section. Customers from outside arrive at the G-Section according to a
Poisson process at a mean rate of 10/hour and they reach the P-section at a mean rate of 2/hour.
The service times at both the sections are exponentially distributed with parameters 15 and 12
respectively. On finishing the job in the G-section, a customer is equally likely to go the P-
section or to leave the store, whereas a customer on finishing his job in the P-section will go to
the G-section with probability 0.25 and leave the store otherwise. Assuming that there is only
one salesman in each section, find the probability that there are 3 customers in the G-section and
2 customers in the P-section. Find also the average number of customers in the store and the
average waiting time of a customer in the store.
Solution: This system given is a Jacksons open queuing system.
1
total arrival rates of
1
S
2
total arrival rates of
2
S
Jacksons flow balance equations for this open model are
1
2
1 1 11 21
1
1 2 21
, 1, 2,.....
10 10
1 1
10 0
4 4
k
j j i ij
i
i i i i
i
r P j k
P P P
P



=
=
= + =
= + = + +
(
= + + =
(

| | ( )
| | ( )
| |
1 2 11
2
2 2 1 12 2 22
1
2 1 22
1 1 1
1
1
1
2
1
10 0 ....... 1
4
2 2
1
2 0 ....... 2
2
1 1 2 1
10 2 10
4 2 4 8
7 42
8 4
7
42
2
42 2
12
7
1
2 12 8
2
i i
i
P
P P P
P

=
= + =
= + = + +
= + =
(
= + + = + +
(

=
=

= =
= + =

Given
1 2
1 2
1 2
12 4 8 2
15, 12, ,
15 5 12 3



= = = = = =
Non-Markovian Queues and Queue Networks
24
P(n
1
customers in S
1
and S
2
customers in S
2
)
( )
( )
( ) ( )
1 2
1 1 2 2
1 2
1 2 2
3 2
1 2
1 1 2 2
, 1 1
4 4 2 2
3, 2 1 1
5 5 3 3
64 1 4 1 256
0.0152
125 5 9 3 16875
12 8
4 2 6
15 12 12 8
1 1 1
6
12 2
n n
P n n
P
L
E W L

| | | | | | | |
=
| | | |
\ . \ . \ . \ .
| | | || | | |
=
| | | |
\ . \ .\ . \ .
| || || || |
= = =
| | | |
\ .\ .\ .\ .
= + = + = + =

= = = | | 10 2 12 = + =
1
2
= hours
1
60
2
= minutes 30 = minutes.
Problem: 23 Consider two servers. An average of 8 customers per hour arrive from outside at
server 1 and an average of 17 customers per hour arrive from outside at server 2. Inter arrival
times are exponential. Server 1 can serve at an exponential rate of 20 customers per hour and
server 2 can serve at an exponential rate of 30 customers per hour. After completing service at
server 1, half of the customers leave the system, and half go to server 2. After completing service
at server 2,
3
4
of the customers complete service, and
1
4
return to server1.
(i) What fraction of the time is server 1 idle?
(ii) Find the expected number of customers at each server.
(iii) Find the average time a customer spends in the system.
(iv) How would the answers to parts (i) (iii) change if server 2 could serve only an average of
20 customers per hour?
Solution: Given
1
8 r =
2
12 21
11 22
17
1/ 2 1/ 4
0
r
P P
P P
=
= =
= =
Jacksons flow balance equations for this open model are
| |
1
2
1 1 1
1
1 11 2 21
2 11
, 1, 2...,
8
1
8 0 0
4
k
j j j ij
i
i i
i
r P j k
r P
P P
P

=
=
= + =
= +
= + +
= + + =

Non-Markovian Queues and Queue Networks

25
( )
| |
( )
( )
1 2
2
2 2 2
1
1 12 2 22
1 22
2 1
1 1
1
1
8 ..... 1
4
17
1
17 0 0
2
1
17 ..... 2
2
1 1
1 8 17
4 2
17 1
8
4 8
i i
i
r P
P P
P

=
= +
= +
= + +
| |
= + + =
|
\ .
= +
(
= + +
(

= + +

( )( )
( ) ( )
1 1
1
1
2
1 49
8 4
7 49
8 4
49 8
7 2 14
4 7
1
2 17 14 17 7 24
2

=
=
| || |
= = =
| |
\ .\ .
= + = + =
(i) Server 1 may be treated as an M/M/1/GP/ / system with 14, 20 0.7

= = = =
0
1 1 0.7 0.3 P = = =
Thus server 1 is idle 30% of the time.
(ii)
1 2
1 1 2 2
L


= +

14 24
20 14 30 24
14 24 7 19
4
6 6 3 3
= +

= + = + =
(iii)
| |
1 2
1 1 19 19
8 17 25
25 3 75
W L r r

| |
= = = = + = + =
|
\ .

2 2
20 24 = < = . So no steady state solution exists.