Professional Documents
Culture Documents
Madhu Jain
Indian Institute of Technology Roorkee
Abstract
This paper presents a study of controllable Markovian queueing system with interdependent rates. The customers behavior is
incorporated according to which balking and reneging with certain probability is taken into account. The models with finite
capacity (FCM) and finite population (FPM) are developed. We derive queue size distribution, which is further employed to
derive formulae for average number of customers in the system and the expected waiting times for both models. Some earlier
results are deduced by setting suitable parameters.
Keywords: Controllable Queue, Interdependent rates, Balking, Reneging, Multi- servers, Queue size
________________________________________________________________________________________________________
I. INTRODUCTION
Markovian analysis is a way of analyzing the current movement of some variables in an effort to forecast its future movement.
Multi-server Markovian models can play a significant role in day-to-day queueing situations encountered, marketing, production,
transportation, computer, communication and manufacturing systems etc. In many real life queueing situations due to long queue
of customers, the arriving customers may be discouraged. The balking and reneging phenomena arise in the queueing system
when the customers leave the system before joining the queue and depart after joining the queue without getting service due to
impatience, respectively. Many researchers have done a lot of work on queueing models with balking and reneging in different
frame-works. Haghighi et al. [6] obtained the steady-state distribution for multi-server Markovian queueing system with balking
and reneging. Abou-El-alta and Hariri [1] discussed M/M/C/N queue with balking and reneging. Various aspects of balking and
reneging can be found in the textbook by Hillier and Lieberman [7]. The single-server Markovian overflow queue with balking,
reneging and an additional server for longer queues was also analysed by Abou-El-alta and Shawky [2]. Jain and singh [10]
derived steady state probability distribution and various characteristics for M/M/m queue with balking, reneging and additional
servers. Abou-El-alta and Kote [3] included the concept of linearly dependent service rate for the M/M/1/N queue with genral
balk function, reflecting barrier, reneging and an additional server for longer queues. The controllable queue with balking and
reneging. To reduce the balking behavior of the customers in the controllable queue, the provision of additional removable
servers, considered by Jain and Sharma (14). Singh et al. (15) have examined a Single server interdependent queueing model
with controllable arrival rates and reneging. Jain et al. (16) described the controllable and interdependent rates for the machine
re-pair problem (MRP) with additional repairman and mixed standbys. Yang et al.(17) considered the optimization and
sensitivity analysis for controlling the arrivals in the queueing system with single working vacation. A Balking and reneging in
multi-server Markovian queuing system was examined by Choudhury and Medhi (18). Mandelblaum and Momcilovic (19)
described a queues with many Servers and Impatient Customers. Kumar and Sharma (20) have present a multi-server Markovian
Feedback Queue with Balking Reneging and Retention of Reneged Customers.
In this investigation we develop finite controllable Markovian queueing model with balking and reneging by applying birthdeath process. We obtain queue size distribution, the average number of customers in the system and waiting time. We also
deduce some particular cases by setting suitable parameters.
queue size reduces to backward threshold level (say r), the arrival rate changes to and the same process continues to be
repeated. The system said to level 0(zero) and 1 when customers arrive with rate and 1 respectively.
The customers may balk depending upon the queue size. The probability of joining the queue is a non-linear function of the
number of servers per customer.
36
The state-dependent arrival rate for finite capacity model (FCM) is given by
e
,
n C
C
(n)
C n R 1
for level 0
e ,
n
C
r R n K,
for level 1
1 e ,
n 1
where K is the finite capacity of the system.
The state-dependent arrival rate for finite population model (FPM) is given as follows
M n e
,
C
( n ) M n
C
e ,
n 1
n
r
1 e ,
n 1
(n)
(1)
n C
n R 1
for level 0
R n M,
(2)
for level 1
n e
C e n C
nC
C n L
(3)
where L takes value K and M for FCM and FPM models, respectively.
(4)
0 e n e Pn 0 e Pn 1 0 n 1 e Pn 1 0 , 1 n C
0 n n Pn 0 n 1 Pn 1 0 n 1 Pn 1 0 , C n r 1
0 r r Pr 0 r 1 Pr 1 0 r 1 Pr 1 0 r 1 Pr 1 1
(5)
(6)
(7)
0 n n Pn 0 n 1 Pn 1 0 n 1 Pn 1 0 , r 1 n R 2
(8)
37
0 R 1 R 1 PR 1 0 R 2 PR 2 0
(9)
0 r 1 r 1 Pr 1 1 r 2 Pr 2 1
(10)
'
'
0 n n Pn 1 n 1 Pn 1 1 n 1 Pn 1 1 , r 2 n R 1
0 P 1
P 1
P 0
P 1
'
'
R n K
0 n n Pn 1 n 1 Pn 1 1 n 1 Pn 1 1 ,
'
'
where n
n 1
'
e ; n
'
R 1 R 1
n 1
R 1 R 1
1 e
(11)
(12)
R 1 R 1
(13)
and n C e n C .
The expressions for Pn ( 0 ) ( 0 n C ) is derived using equations (4) & (5) and is given by,
n
1 e
P0 0 , 0 n C
Pn ( 0 )
n! e
(14)
The expression for Pn ( 0 ) ( C n r ) are recursively obtained from equation (6) as,
n
n c 1
1
e (C !)
.C
Pn ( 0 )
P0 0 ,
n
e
( n! )
Bi
Bi 1
where
C n r
(15)
i C 1
i C
C e
e ( C ! ) 1 .C n c 1
Pn ( 0 )
P0 0 n r Pr 1 1 , r 1 n R 1
n
e
( n! )
Bi
(16)
i C 1
where 1 1 ; 2 2 ; i i i 1 i i 2 , for i 3 ; i
r i 1 r i 1
r i
and i
r i 2
Using equation (9), the expression for Pr 1 (1) in terms of P0 0 is given by,
r i
Pr 1 (1) W . P0 0
e
W
e
where
(17)
(C !)
( R!)
R c 1
.C
R
B i . R r
i C 1
Pn ( 0 )
P0 0 n r W . P0 0 , r 1 n R 1
n
e
( n! )
Bi
(18)
i C 1
Pn (1) n rW .P0 ( 0 ) ,
'
r 1 n R
(19)
r i 1 r i 1
'
'
'
'
'
'
'
'
r i
r i2
'
'
where 2 2 ; i i i 1 i i 2 , for i 3 ; i
and
'
r i
Pn 1 [ n rW D n ] P0 0 ,
'
R n M
(20)
38
where D R 1 R 1 R 1 r .W R 1S ; D n n D n 1 n D n 2 , for n R 2 ;
'
and S
e
'
R 1
(C !)
.C
'
'
R 1 c 1
( R 1! )
R 1
Bi
i C 1
Pn (1) 1
Pn ( 0 )
n0
(21)
n r 1
so that we obtain
P0 ( 0 )
1 e
n 0 n! e
C
n C 1 e
r
(C !)
( n! )
.C
n c 1
n
Bi
i C 1
n r 1 e
R 1
(C !)
( n! )
.C
n c 1
n
(22)
P0 0 n r W
Bi
i C 1
R
n rW
'
n r 1
n rW D n
'
n R 1
e
( n 1)! e
n0
C
n C 1
(C !)
.C
(( n 1)! )
n c
Bi
i C 1
1
n
n r 1
n
e ( C ! ) 1 .C n c 1
n n rW
n
e
(( n 1)! )
Bi
R 1
i C 1
n n r W
K
n n rW
'
n r 1
'
Dn
P0 (0 )
(23)
n R 1
where is the effective mean arrival rate of the system and is given by
R 1
n0
Pn ( 0 )
C
n 1 Pn ( 0 )
n C 1
C
n 1 1 Pn (1)
n r 1
(25)
0 M n e n e Pn 0 M ( n 1) e Pn 1 0 n 1 e Pn 1 0
1 n C
0 n n Pn 0 n 1 Pn 1 0 n 1 Pn 1 0 , C n r 1
(26)
(27)
39
0 r r Pr 0 r 1 Pr 1 0 r 1 Pr 1 0 r 1 Pr 1 1
(28)
0 n n Pn 0 n 1 Pn 1 0 n 1 Pn 1 0 , r 1 n R 2
(29)
0 R 1 R 1 PR 1 0 R 2 PR 2 0
(30)
0 r 1 r 1 Pr 1 1 r 2 Pr 2 1
(31)
'
'
0 n n Pn 1 n 1 Pn 1 1 n 1 Pn 1 1 , r 2 n R 1
0 P 1
P 1
P 0
P 1
'
'
R n M
0 n n Pn 1 n 1 Pn 1 1 n 1 Pn 1 1 ,
'
'
'
C
where n M n
n 1
R 1 R 1
R 1 R 1
(32)
(33)
R 1 R 1
C
'
e ; n M n
n 1
(34)
1 e
and n C e n C
Equation (25) provides,
M e
P1 0
P 0
e 0
(35)
P 0 , 1 n C
Pn ( 0 )
M n ! n! e 0
M!
(36)
Pn ( 0 )
P0 0 , C n r
n
M n ! e
( n! )
Bi
M!
i C 1
i C
.
C e
where B i 1
(37)
From equations (28) and (29), the expression for Pn ( 0 ) ( r 1 n R 1 ) is obtained as,
Pn ( 0 )
n
n c 1
1
e (C !)
.C
P0 0 n r Pr 1 1 , r 1 n R 1 (38)
n
M n ! e
( n! )
Bi
M!
i C 1
where 1 1 ; 2 2 ; i i i 1 i i 2 , for i 3 ; i
and i
r i 1 r i 1
r i
r i 2
r i
Pr 1 (1) U .P0 0
where U
M n ! e
M!
(C !)
(39)
1
( R!)
.C
R c 1
B i . R r
i C 1
Substituting the value of Pr 1 (1) from (39) in equation (40), Pn ( 0 ) is obtained as,
40
Pn ( 0 )
P0 0 n r U . P0 0 , r 1 n R 1
n
M n ! e
( n! )
B
i
M!
(40)
i C 1
Pn (1) n rU .P0 ( 0 ) , r 1 n R
'
(41)
r i 1 r i 1
'
where 2 2 ; i i i 1 i i 2 , for i 3 ; i
'
'
'
'
'
'
'
'
r i
r i2
'
and
'
r i
Pn 1 [ n rU H n ] P0 0 ,
'
R n M
(42)
where H n n H n 1 n H n 2 , n R 2 ; H R 1 R 1 R 1 r .U R 1G
'
and
'
M ( R 1) ! e
M!
R 1
(C !)
'
'
R 1 c 1
.C
( R 1! )
R 1
Bi
i C 1
P0 ( 0 )
M n ! n! e
M!
n0
e ( C ! ) 1 .C n c 1
n C 1 M n ! e
( n! )
Bi
r
M!
i C 1
R 1
n ! e
M!
n r 1 M
(C !)
( n! )
.C
n c
n rU
Bi
i C 1
n rU
'
n r 1
( n r U H n )
'
(43)
n R 1
M n ! ( n 1)! e
M!
n0
n C 1
M! e
M n ! e
n
(C !)
.C
(( n 1)! )
n c
Bi
i C 1
n
n c 1
1
1
M ! e (C !)
.C
n
n n rU
n r 1 M n ! e
(( n 1)! )
Bi
i C 1
R 1
n r 1
n n rU
'
n ( n r U H n ) P0 ( 0 )
'
(44)
n R 1
41
1) Case I: When 0 i.e. there is no balking, the performance indices for finite capacity and finite population models
reduce to the following forms;
ForFP Mmodel:
1 e
P0 0 ,
n! e
n
1
e
Pn ( 0 )
P0 0 ,
n
e C n c C B
i
i C 1
n
1
e
P0 0 Q n r V . P0 0 ,
n
e
n c
C
C ! Bi
i C 1
1 n C
C n r
r 1 n R 1
Where Q1 1 ; Q 2 2 ; Q i i Q i 1 i Q i 2 , for i 3 ; i
r i 2
and V
e
r i
1
R
R c
C
C ! B i .Q R r
(45)
r i 1 r i 1
r i
i C 1
'
Q n rV
. P0 ( 0 )
,
r 1 n R
Pn (1)
'
( Q n r V A n ) P0 0 , R n K
(46)
r i 1 r i 1
'
where Q 2 2 ; Q i i Q i 1 i Q i 2 , for i 3 ; i
'
'
'
'
'
'
'
'
r i
r i2
'
and
also
and
'
r i
An n An 1 n An 2 ; AR 1 R 1Q R 1 r .V R 1T , for n R 2
R 1
.
e
1
'
'
R 1 c
C!
'
'
R 1
Bi
i C 1
e
M!
P0 0 ,
M n ! n!
e
n
e
M!
1
Pn ( 0 )
P0 0 ,
e
n
n
!
C !C
Bi
i C 1
n
M
!
e
1
P0 0 n r H . P0 0
n
M n !
e C !C n c B
i
i C 1
1 n C
(47)
C n r
,r 1 n R 1
42
H
M n ! e
M!
1
R
R c
C
C ! B i . R r
r i 1 r i 1
i i i 1 i i 2 , for i 3 ;
2 2 ;
1 1 ;
where
r i
r i 2
r i
and
i C 1
And
H . P0 ( 0 ) ,
r 1 n R
Pn (1) n r
'
( n r H E n ) P0 0 , R n M
'
(48)
r i 1 r i 1
'
'
'
'
'
'
'
'
r i
'
Also E n n E n 1 n E n 2 , n R 2 ; E R 1 R 1 R 1 r H R 1 Z
'
'
and Z
M ( R 1) ! e
M!
R 1
'
1
R 1
R 1 c
C
C ! Bi
r i2
'
'
where 2 2 ; i i i 1 i i 2 , i 3 ; i
'
r i
,.
i C 1
2) Case II: When C=1, 0 , 0 , i. e. for single server queue without discouragement,
We set
i
for i R
1
for r i R and , i R L
In this case our model reduces to the M/M/1 interdependent queueing model with controllable arrival rates. The same case for
infinite capacity was discussed in [9]. We get results for FCM and FPM models as follow;
FCM Model:
P0 0 ,
1 n 1
e
n
1
e
Pn ( 0 )
P0 0 ,
1 n r
n
e ( n! ) B
i
i2
n
1
e
P0 0 q n r a . P0 0 , r 1 n R 1
n
e
(
n
!
)
B
i2
where
1
( R!)
(49)
B i .q R r
i2
And
q ' a . P ( 0 ) ,
0
Pn (1) n r
'
( q n r a d n ) P0 0 ,
r 1 n R
(50)
R n M
r i 1 r i 1
'
'
'
'
'
'
'
'
r i
d n n d n 1 n d n 2 , n R 2 ; d R 1 R 1q R 1 r .a R 1 f
'
'
'
'
r i2
'
'
where q 2 2 ; q i i q i 1 i q i 2 , for i 3 ; i
'
r i
43
and f
R 1
1
(( R 1)! )
R 1
Bi
i2
FPM Model:
In this model, we obtain
M e P 0 ,
0
e
n
e
M!
1
Pn ( 0 )
P0 0 ,
n
M n ! e ( n! ) B
i
i2
n
e
M!
1
P0 0
n
M n ! e
( n! ) B i
i2
where
1 1;
2 2 ;
'
'
l
M n ! e
M!
'
i
'
1
( R!)
'
i 1
i
'
'
n 1
'
n r l . P0
for i 3 ;
i2,
(51)
1 n r
0 , r
'
1 n R 1
r i 1 r i 1
r i
i
'
r i 2
r i
and
B i .
'
Rr
i2
Also
l .P (0 ) ,
Pn (1) n r 0
"
( n r l h n ) P0 0 ,
''
r 1 n R
(52)
R n M
r i 1 r i 1
'
where 2 2 ; i i i 1 i i 2 , for i 3 ; i
"
''
"
'' ''
'' ''
''
hn n hn 1 n hn 2 , n R 2 ; hR 1 R 1
"
"
and p
M ( R 1) ! e
M!
R 1
"
'
R 1 r .l
1
(( R 1)! )
R 1
r i
R 1 p
"
r i2
'
"
r i
Bi
i C 1
3) Case III: When e=0, 0 , 0 and 1 , then our model reduces to the classical finite population M/M/C model.
4) Case IV: When 0 , 0 i.e. the customers behavior is not taken account and N , our results of FCM model
tally with those of Aftab and Maheswari [11].
V. DISCUSSION
In this investigation we have developed finite controllable Markovian models with balking and reneging. The arrival rate
influenced by the queue size is included which is further controlled by forward and backward preassigned threshold levels. The
explicit formulae for queue size distribution, average number of customers in the system and expected waiting times are
established. The concept of balking and reneging used in the system modeling makes our models closer to the real life situation.
REFERENCES
[1]
Abou-El-alta, M.O. and Hariri, A.M.A. (1992): The M/M/C/N queue with balking and reneging. Computers and Operations Research, Vol. 19(8), pp. 713716.
44
Abou-El-alta, M.O. and Shawky, A.I. (1992): The single-server Markovian overflow queue with balking, reneging and an additional server for longer
queues. Microelectron. Reliab., Vol. 32, pp. 1389-1394.
Abou-El-alta, M.O. and Kote, K.A.M. (1992): A linearly dependent service rate for the queue M/M/1/N with genral balk function reflecting barrier,
reneging and an additional server for longer queues. Microelectron. Reliab., Vol. 32(12).
Aftab Begum, M.I. and Maheswari, D. (2002): The M/M/C interdependent queueing model with controllable arrival rates, OPSEARCH, Vol. 39(2) pp. 89110.
Gross and Harris, C.M. (1974): Fundamental of Queueing Theory, John Wiley, New York.
HAGHIGHI, A.M., Medhi, J. and Mohanty, S.G. (1986): On multi-server Markovian queueing system with balking and reneging. Computers and
Operations Research, Vol. 13, pp. 421-425.
Hillier, F. S. and Lieberman, G.J. (1992): Operations Research. Holden-Day Inc: San Francisco.
Jain, M. (1998): Finite population loss and delay queueing system with nopassing. OPSEARCH, Vol. 35(3), pp. 261-276.
Jain, M. (1998): M/M/m queue with discouragement and additional servers. Gujarat Statistical Review, Vol. 25(1-2), pp. 31-42.
Jain, M. and singh, P. (2002): M/M/m queue with balking and reneging and additional servers. International Journal of Engineering.
Srinivasa Rao, K., Shobha, T. and Srinivasa Rao, P. (2000): The M/M/1 interdependent queueing model with controllable arrival rates, OPSEARCH, Vol.
37(1).
Washurn, A. (1974): A multi-server queue with nopassing. Oper. Res., Vol. 16, pp. 428-834.
Ma Z. Y et al.,(2007): discrete time queueing system with two classes of customers and priorities. Sys. Engg. Theo. Pra. Vol. 27, pp.91-98.
Jain M. and Sharma G.C.(2004): The controllable queue with balking and reneging. To reduce the balking behavior of the customers in the controllable
queue, the provision of additional removable servers, Nepali Math. Sci. Report Vol.22, pp.113-120.
Singh C.J et al. (2007): Single server interdependent queueing model with controllable arrival rates and reneging Pak. J. Statist. Vol.23, pp. 171-178.
Jain M.et al. (2009): The controllable and interdependent rates for the machine re-pair problem (MRP) with additional repairman and mixed standbys Raj.
Acad. Phy. Sci. Vol.8, pp. 447-456.
Yang D.Y. et al.(2010): The optimization and sensitivity analysis for controlling the arrivals in the queueing system with single working vacation J.
Comput. App. Math. Vol.234,pp. 545-556.
Choudhury, A. and Medhi, P. (2011), Balking and reneging in multiserverMarkovian queuing system, International Journal of Mathematics in Oper. Res.,
Vol. 3, pp. 377-394.
Mandelblaum, A. and Momcilovic, P. (2012) Queues with Many Servers and Impatient Customers, Mathematics of Operations Research, Vol. 37,pp. 14165.
20. Kumar,R.and Sharma, S.K.(2014): A Multi-Server Markovian Feedback Queue with Balking Reneging and Retention of Reneged Customers,
Advanced Modeling and Optimization (AMO), Vol.16(2), pp.395-406.
45