OR372-Dr.

Khalid Al-Nowibet
1
4.4 M/M/s/k Queueing Model
Characteristics
1. Interarrival time is exponential with rate
Arrival process is Poisson Process with rate
2. Interarrival time is exponential with rate
Number of services is Poisson Process with rate
3. Multiple Server:
Number of servers = s
Independent servers
Parallel service channels
Identical servers
Random choice of idle servers
4. System size is finite : Total number in system k
5. Queue Discipline : FCFS
Notation
M / M / s / k / FCFS
OR372-Dr.Khalid Al-Nowibet
2
4.4 M/M/s/k Queueing Model
Steady-State Distribution
State of the system
system is in state n if there are n customers in the
system (waiting or serviced)
Let P
n
be probability that there are n customers in the
system in the steady-state. n = 0 , 1 , 2 , 3 , ,k
OR372-Dr.Khalid Al-Nowibet
3
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Rate Diagram:
Consider an M/M/3/k system
system in state n = 1 1 server busy total service speed =
system in state n = 2 2 servers busy total service speed = 2
system in state n = 3 3 servers busy total service speed = 3
system in state n = 4 3 servers busy total service speed = 3

system in state n > 3 3 servers busy total service speed = 3

Queue is closed
system size = k
1
2
3

E
x
p
o
n
e
n
t
i
a
l

S
e
r
v
i
c
e

e
OR372-Dr.Khalid Al-Nowibet
4
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Rate Diagram:
1. Arrival rate = for all states of the system
2. Service rate = n if n s
Service rate = s if s< n k
0 1

2
. . .

3
s-1

(s-1)
s

s
s+1
. . .
s
s

k-1 k
. . .

s s
OR372-Dr.Khalid Al-Nowibet
5
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Balance Equations:
n = 0 P
0
= P
1
n = 1 P
1
+ P
1
= P
0
+ 2P
2
(+)P
1
= P
0
+ 2P
2
n = 2 P
2
+ 2P
2
= P
1
+ 3P
3
(+2)P
2
= P
1
+ 3P
3
n = 3 P
3
+ 3P
3
= P
2
+ 4P
4
(+3)P
3
= P
2
+ 4P
4
. . . . . . . . . . .
n = s P
s
+ sP
s
= P
s-1
+ sP
s+1
(+s)P
s
= P
s-1
+ sP
s+1
Average
Rate in to
State n
Average
Rate out of
State n
=
0 1

2
. . .

3
s-1

(s-1)
s

s
s+1
. . .
s
s

OR372-Dr.Khalid Al-Nowibet
6
4.4 M/M/s/k Queueing Model
Average
Rate in to
State n
Average
Rate out of
State n
=
Steady-State Distribution
Balance Equations:
k-1 k
. . .

s s
. . . . . . . . . . .
n = s P
s
+ sP
s
= P
s-1
+ sP
s+1
(+s)P
s
= P
s-1
+ sP
s+1
n = s+1 P
s+1
+ sP
s+1
= P
s
+ sP
s+2
(+s)P
s+1
= P
s
+ sP
s+2
. . . . . . . . . . .
n = k sP
k
= P
k-1
sP
k
= P
k-1
OR372-Dr.Khalid Al-Nowibet
7
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Eq-1: P
0
= P
1
Eq-2: (+)P
1
= P
0
+ 2P
2
Eq-3: (+2)P
2
= P
1
+ 3P
3
. . . . . . . . . . .
Eq-n: (+n)P
n
= P
n-1
+ (n+1)P
n+1
n < s
. . . . . . . . . . .
Eq-n: (+s)P
n
= P
n-1
+ sP
n+1
n s
. . . . . . . . . . .
Eq-k: sP
k
= P
k-1
OR372-Dr.Khalid Al-Nowibet
8
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Eq-1 P
0
= P
1
Eq-2 P
1
= 2P
2
Eq-3 P
2
= 3P
3
. . . . . . . . . . .
Eq-n P
n-1
= nP
n
,n < s
. . . . . . . . . . .
Eq-n P
n-1
= sP
n
, n s
. . . . . . . . . . .
Eq-k P
k-1
= sP
k
P
1
= (/) P
0
P
2
= (/2)P
3
P
3
= (/3)P
2
. . . . . .
P
n
= (/n)P
n-1
. . . . . .
P
n
= (/s)P
n-1
. . . . . .
P
k
= (/s)P
k-1
OR372-Dr.Khalid Al-Nowibet
9
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Make all equations functions of P
0
only:
Eq-1 P
1
= (/) P
0
Eq-2 P
2
= (/2)P
1
Eq-3 P
3
= (/3)P
2
. . . . . . . . . . .
Eq-n P
n
= (/n)P
n-1
,n < s
. . . . . . . . . . .
Eq-n P
n
= (/s)P
n-1
, n s
. . . . . . . . . . .
Eq-k P
k
= (/s)P
k-1
P
1
= (/) P
0
P
2
= (
2
/2
2
)P
0
P
3
= (
3
/6
3
)P
0
. . . . . .
P
n
= (
n
/n!
n
)P
0
. . . . . .
P
n
= [
n
/(s! s
n-s

n
)]P
0
. . . . . .
P
k
= [
k
/(s! s
k-s

k
)]P
0
OR372-Dr.Khalid Al-Nowibet
10
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Make all equations functions of P
0
only:
Let
Eq-1 P
1
= P
0
Eq-2 P
2
= (
2
/2)P
0
Eq-3 P
3
= (
3
/6)P
0
. . . . . .
Eq-n P
n
= (
n
/n!)P
0
n < s
. . . . . .
Eq-n P
n
= [
n
/(s! s
n-s
)]P
0
n s
. . . . . .
Eq-k P
k
= [
k
/(s! s
k-s
)]P
0

=
OR372-Dr.Khalid Al-Nowibet
11
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Computing P
0
:
P
0
+ P
1
+ P
2
+ +P
s
+P
s+1
+ + P
n
+P
n+1
+ = 1
P
0
+ P
0
+ (
2
/2) P
0
+ (
3
/3!)P
0
+ + (
s
/s!)P
0
+ + (
k
/s! s
k-s
)P
0
= 1
P
0
[ 1 + + (
2
/2) + (
3
/3!) + + (
s
/s!) + + (
k
/s! s
k-s
) ] = 1
P
0
= [ 1 + + (
2
/2) + (
3
/3!) + + (
s
/s!) + + (
k
/s! s
k-s
) ]
1
1 P
n
=

n
OR372-Dr.Khalid Al-Nowibet
12
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
Computing P
0
:
1
1
0
0
! !
P

=

=

k
s n
s n
n
s
n
n
s s n
finite sum finite value
P
n
is a function of P
0
if P
0
= 0 system is infinite
system is finite if an only if P
0
> 0

s s
< 1
No Steady-State Condition on , , and s
finite sum finite value
P
0
> 0 and finite always for any , and s
NO
OR372-Dr.Khalid Al-Nowibet
13
4.4 M/M/s/k Queueing Model
Steady-State Distribution
Solution of Balance Equations:
P
n
Formula:
0
P
!
P
n
n
n
n

=
0
P
!
P
n s n
n
n
s s

=

n < s
s n k
0
P
!
P
n
n
n

=
0
P
!
P
s n
n
n
s s

=
n < s
snk
1
1
0
0
! !
P

=

=

k
s n
s n
n
s
n
n
s s n
OR372-Dr.Khalid Al-Nowibet
14
4.4 M/M/s/k Queueing Model
Performance Measures
In steady state

e
,
b
, , P
0
L
B
= E[busy servers] = E[#Cust. in service]
L
s
= L
q
+ L
B
W
s
= W
q
+ (1/)
L
s
=
e
W
s
L
q
=
e
W
q
L
B
=
e
W
B
BP = Blocking Probability
Know 4 measures all measures are known
System is
in Steady Stead
OR372-Dr.Khalid Al-Nowibet
15
4.4 M/M/s/k Queueing Model
Performance Measures
1. Blocking Probability BP:
BP = Pr{system is full} = P
k
2. Effective Arrival Rate
e
:

e
= arrival rate . Pr{system is not full}
= . Pr{ n k }
= . (1 BP) = . (1 P
k
)
OR372-Dr.Khalid Al-Nowibet
16
4.4 M/M/s/k Queueing Model
Performance Measures
3. Average Customers in System L
s
:

=
=
0 n
n s
P n L

=

=
+ =
k s
s n
n
1
0 n
n
P n P n

=

=

=
k
s n
n n
s s s
s n
0 0
1 - s
0 n
P
!
n P
!
n

+ =

= =
k
n
n
s n
s
s
s
s n
1 - s
0 n
0
n n
!
P
OR372-Dr.Khalid Al-Nowibet
17
4.4 M/M/s/k Queueing Model
Performance Measures
3. Average Busy servers L
B
:
L
B
= E[busy servers] = E[#Cust. in service]
L
B
= 0.P
0
+1.P
1
+ 2.P
2
+ + s.P
s
+ s.P
s+1
+s.P
s+2
+s.P
s+3
+
L
B
= 0.P
0
+1.P
1
+ 2.P
2
+ + s ( Pr{ n s })
L
B
= 0.P
0
+1.P
1
+ 2.P
2
+ + s (1 Pr{ n < s })

+ =

=

=
k s
s n
n
1
0 n
n B
P 1 s P n L
OR372-Dr.Khalid Al-Nowibet
18
4.4 M/M/s/k Queueing Model
Performance Measures
4. Average Busy servers L
B
:
L
B
= E[busy servers] = E[#Cust. in service]
Using Littles Formula: L
B
=
e
W
B
W
B
= Average time server is busy
= Expected service = E[S] = 1/
L
B
=
e
W
B
=
e
/
OR372-Dr.Khalid Al-Nowibet
19
4.4 M/M/s/k Queueing Model
Performance Measures
5. Average Customers in Queue L
q
:
L
q
= L
s
L
B
or
Number of customer in queue = 0 with Pr{n s} = P
0
+P
1
+P
2
++P
s
Number of customer in queue = 1 with Pr{n = s+1} = P
s+1
Number of customer in queue = 2 with Pr{n = s+1} = P
s+2
.
L
q
= 0.(P
0
+P
1
+P
2
++P
s
)+ 1.P
s+1
+ 2.P
s+2
+ 2.P
s+3

= L
s

e
/

=
=
s n
n
P s) (n
OR372-Dr.Khalid Al-Nowibet
20
4.4 M/M/s/k Queueing Model
Performance Measures
6. Utilization of the System U:
U = Pr{ n > 0 } = P
1
+ P
2
+ P
3
+ + P
k
= 1 P
0
7. Utilization of the Service SU:
SU = Pr{ all servers busy } = Pr{ n s }
SU = 1 (P
0
+ P
1
+ P
2
++ P
s1
)
OR372-Dr.Khalid Al-Nowibet
21
4.4 M/M/s/k Queueing Model
Performance Measures
8. Average Time Spent in System W
s
:
L
s
=
e
.W
s

9. Average Waiting time in Queue W
q
:
L
q
=
e
.W
q

L
s
W
s
=

e
L
q
W
q
=

e
OR372-Dr.Khalid Al-Nowibet
22
4.4 M/M/s/k Queueing Model
Example:
A barber shop has 3 barbers and a total of 7 waiting seats.
Interarrival times are exponentially distributed, and an average of 20
prospective customers arrive each hour at the shop. Those
customers who find the shop full do not enter. The barber takes an
average of 12 minutes to cut each customer's hair. Haircut times are
exponentially distributed.
1. If you arrive to the shop, what is the probability that you get served?
2. On average how many hair cuts is performed per hour?
3. If you enter the shop at 9:00 am when do you expect to leave?
4. What is the probability that a customer waits if he enters?
5. What is the average empty seats in the shop?
6. The owner of the shop decided to accept any arrival to enter the
shop, what is the minimum he must hire?
OR372-Dr.Khalid Al-Nowibet
23
4.4 M/M/s/k Queueing Model
Example:
Arrivals: = 20 cust/hr Poisson
Service: E[S] = 12 min. Exponential
= 1/E[S] = 1/12 cust/min
= 60/12 = 5 cust/hr
Number of Servers: s = 3
System Size = waiting seats + service seats = 7 + 3 = 10
M/M/3/10 /s = 20/(3(5)) = 1.33
OR372-Dr.Khalid Al-Nowibet
24
4.4 M/M/s/k Queueing Model
Example:
1. If you arrive to the shop, what is the probability that you get served?
Pr{get served} = Pr{enter} = Pr{n < 10} = 1 P
10
Pr{get served} = 1 0.2685 = 0.7315
1
1
0
0
! !
P

=

=

k
s n
s n
n
s
n
n
s s n
0
P
!
P
n
n
n

=
0
P
!
P
s n
n
n
s s

=
n < s
snk
n 0 1 2 3 4 5 6 7 8 9 10
P
n
0.0033 0.0133 0.0266 0.0355 0.0473 0.0631 0.0841 0.1121 0.1495 0.1993 0.2658
OR372-Dr.Khalid Al-Nowibet
25
4.4 M/M/s/k Queueing Model
Example:
2. On average how many haircuts is performed per hour?
Average number of hair cuts is performed per hour =
e
= (1 P
10
) = 20 (0.7315) = 14.684 haircut/hr.
3. If you enter the shop at 9:00 am when do you expect to
leave?
E[Departure time] = 9:00 + W
s
W
s
= L
s
/
e
= (nP
n
)/
e
= 7.62/14.684 = 0.5186 hr = 0:32
E[Departure time] = 9:00 + W
s
= 9:00 + 00:32 = 9:32
OR372-Dr.Khalid Al-Nowibet
26
4.4 M/M/s/k Queueing Model
Example:
4. What is the probability that a customer waits if he enters?
Pr{waiting} = Pr{ 3 n < 10}
= P
3
+ P
4
+ . . . + P
9
= 0.691
5. What is the average empty seats in the shop?
E[# empty seats] = waiting seats E[# waiting customers]
L
q
= L
s

e
/ = 7.6154 (14.684)/5 = 4.6786
E[# empty seats] = 7 4.686 = 2.2314 seats
OR372-Dr.Khalid Al-Nowibet
27
4.4 M/M/s/k Queueing Model
Example:
6. The owner of the shop decided to accept any arrival to
enter the shop, what is the minimum barber he must hire?
Accept any arrival system size = k
The new system is M / M /s
/s must be less than 1
= 4 4/s < 1 s > 4
s = 5 , 6 , 7 , . . . minimum number = 5 barabers