IJE TRANSACTIONS C: Aspects Vol. 25, No.

3, (September 2012) 223-230

Intrnational Journal of Engineering TECHNICAL

Journal Homepage: www.ije.ir NOTE

Working Vacation Queue with Second Optional Service and Unreliable Server

M. Jain a, D. Chauhan* b

aDepartment of Mathematics, Indian Institute of Technology, Roorkee-247667

bDepartment of Mathematics, Allenhouse Institute of Technology, Kanpur-208007

ABSTRACT ARTICLE INFO

An M/M/1 queueing system with Article history:

second optional service and Received 2 January 2010
unreliable server is studied. We Accepted in revised form 17 May 2012
consider that the server works at
different rate rather than being idle
during the vacation period. The Keywords:
customers arrive to the system Working Vacation
according to Poisson process with Second Optional Service
Matrix Geometric Technique
state dependent rates depending Server Breakdown
upon the server’s status. All Repair
customers demand the first Queue Size
essential service whereas only some
of them demand the second
optional service. A customer either
may leave the system after the first
essential service with probability
(1-r) or at the completion of the
first essential service go for second
optional service with probability r
(0≤r≤1). The server may breaks
down according to Poisson process
during the busy and working
vacation duration. Both service
times in vacation and in service
period are exponentially distributed.
The matrix geometric technique is
used for the analysis of the
concerned queueing system. The
sensitive analysis is also performed
to examine the variation of the
system performance characteristics
with various input parameters.

doi:
10.5829/idosi.ije.2012.25.03c.06
 *Corresponding Author Email: deepa25july@yahoo.com (D.
Chauhan)
service discipline; the server was assumed to work at
different service rates rather than completely stopping
the service during vacation. A queueing system with c
servers and a threshold type vacation has been
considered by Tian et al. [8]. Ke [9] studied the
operating characteristics of an Mx/G/1 queueing system
under a variant vacation policy. Lin and Ke developed a
cost model to determine the optimal values of the
number of the servers and working vacation rate for
M/M/R queue with vacation.
In many real cases, the server may experience
breakdowns, so that a more realistic queueing model is
that which incorporates the assumption of unreliable
server. Cao et al. [10] investigated an M/G/1 queueing
model with repairable server. Reliability analysis of
M/G/1 queueing system with server breakdowns and
vacations has been examined by Li et al. [11]. Wang et
al. [12] analyzed the retrial queue with server
breakdowns and repairs. Ke [13] studied the N-policy
M/G/1 queue with server vacations, startup and
breakdowns. Almasi et al. [14] examined a single server
retrial queue with finite number of homogeneous
sources of calls and a single removable server. Stability
conditions are provided by Sherman et al. [15] for an
M/M/1 retrial queue with infinite capacity orbit. Mx/G/1
 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230 224

the vacation as well as while working. The paper is system under vacation policies with startup/closedown
organized as follows. In section 2, we outline the times has been examined by Ke [16]. Wang et al. [17]
underlying assumptions and notations to develop considered a single unreliable server in an M x/M/1
mathematical model under study. The governing steady queueing system with multiple vacations. M/G/1
state equations are constructed by taking appropriate queueing system with a single removable and unreliable
transition rates. In section 3, matrix geometric solution server has been considered by Wang et al. [18], by
of the system is given. In section 4, various assuming that the server needs a startup time before
performance characteristics of the system are providing the service. Yang and Alfa [19] studied a multi
formulated explicitly in terms of steady sate server queueing system with identical unreliable server
probabilities. In section 5, we perform comparative with phase type distributed service time.
study of the system characteristics for various input Most studies on queueing models have considered the
parameters. In the last section, conclusions and future main service. However, in many real service systems
scopes of the work are provided. some customers require the main as well as subsidiary
services provided by the server. A pioneer work on such
queueing situation was done by Madan [20], who first
2. MATHEMATICAL MODEL introduced the concept of second optional service while
studying the time dependent as well as steady behavior of
We consider a single server vacation model with an M/G/1 queueing system with no waiting capacity,
second optional service and unreliable server. The using supplementary variable technique. An M/G/1
following assumptions are made to describe the model; queueing system with second optional service has also
v A single server queue begins with a working studied by Madan [20]. Medhi
vacation, when the system is empty. During the [21] proposed an M/G/1 queueing model with second
vacation the customers arrive in Poisson fashion to optional channel and developed the explicit expressions
the system with rate λV. for the mean queue length and mean waiting time.
v The customers are served at mean rate of µ V during Supplementary variable technique was used to develop
the working vacation and the server is prone to the time dependent probability generating function in
breakdown with mean rate αV. When the server terms of their Laplace transform for M/G/1 queue by Al-
breaks down during vacation, it is sent for repair Jararha and Madan [22]. M x/G/1 queueing system with
with rate βV. two phases of heterogeneous service under N-policy was
v When the server is not on working vacations, the examined by Choudhury and Paul [23]. Wang [24]
customers arrive to the system according to Poisson considered an M/G/1 queueing system with second
process with rate λB. optional service and server breakdowns based on
v Two types of services are provided to the supplementary variable technique. Mx/G/1 queueing
customers. The first essential service is needed to system was studied with additional second phase of
all arriving customers with mean rate µ1. As soon as optional service and unreliable server by Choudhury et
the first service of a customer is completed, then al. [25].
with probability r, he may opt for the second service Many authors have paid attention on matrix
or else with probability (1-r), he may opt to leave geometric approach which is used to solve the more
the system. The second service times are assumed complex queueing problems having phase
to be exponentially distributed with mean service
arrivals/services [26]. A matrix geometric approximation
rate µ2. for tandem queues has been examined by Gomez [27]
v Assume that the life time of the server is with blocking and repeated attempts. Matrix geometric
exponentially distributed with mean 1/α1 in first solution of the nested QBD chains was given by Choi et
essential service. In second optional service, the al. [28]. The queue length distribution was computed by
server may fails according to exponential
Gray et al. [29] using matrix geometric method for a
distribution with rate α2. queueing model with multiple types of server
v After breakdown, the server immediately sent for breakdowns. Ke and Wang [30] used matrix geometric
repair. The repair time distributions while server
theory in studying the machine repair problem with R
fails during essential and optional service phases
servers who take vacation of random length. A single
are exponentially distributed with mean rate β1 and server working vacation queueing model with multiple
β2, respectively. types of server breakdown via matrix geometric approach
v The server immediately starts to serve the was described by M. Jain and A. Jain [31].
customers after it is fixed. In this paper, we develop a working vacation
For mathematical formulation purpose, we define queueing model with second optional service and
the following steady state probabilities: unreliable server. The server may breaks down during
P(0,n,V) Probability that there are n \ customers in the
system when the server is on working
vacation
225 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230

P(1,n,V) Probability that there are n customers in the éB00 B

01
ê A A ú

system when the server is in broken down ê

B
10 1 0 ú

state during working vacation ê A2 A1 A0

ê ú

ê --- ú

P(1,n,B) Probability that there are n customers in the Q=ê ---

ú

system when the server is rendering first ê ú

ê A2 A1 A0 ú

essential service ê A A A ú

ê 2 1 0 ú

P(2,n,B) Probability that there are n customers in the ê - - -ú

ê ú

system when the server is rendering second ë û

optional service The matrix can be decomposed in to sub matrices B00,

P(3,n,B) Probability that there are n customers in the B01, B10, A0, A1, A2 as follows:
system and the server is in broken down state B = [- (lV + lB )];
00
while rendering first essential service
P(4,n,B) Probability that there are n customers in the B01 = [ln 0 l
B
0 0 0]
system and the server is in broken down state
m
while rendering second optional service é V ù
ê 0 ú
The steady state equations governing the model are ê ú
constructed as follows: B =ê ê(1
- r )m
1 ú
ú
10 ê ú
m2
( lV + l B ) P (0 ,0 ,V ) = (1) ê 0 ú
mV P (0 ,1,V ) + (1 - r ) m1 P (1,1, B ) + m 2 P ( 2,1, B ) ê ú
ê 0 ú

ë û
(2) 0 0 0 0 0ù
(lV + aV + mV )P(0, n,V ) = él V
m P(0, n + 1,V ) + l P (0, n - 1,V ) + b P (1,1,V ), n > 0 ê 0
ê
l V
0 0 0 0 ú ú

V V V ê 0 0 lB 0 0 0 ú
A0 = ê ú
ê 0 0 0 lB 0 0 ú
(lV + bV ) P(1,1,V ) = aV P (0,1,V ) (3) ê 0 0 0 0 0 ú
ê lB ú
ê 0 0 0 0 0 ú
ë l
(lV + bV )P(1, n,V ) = aV P(0, n,V ) + lV P(1, n - 1,V ), n > 1 (4) Bû
é-(lV +aV + mV ) aV 0 0 0 0 ù
ê 0 0 0 0 ú
ê bV - (lV + bV ) ú
(a1 + lB + (1 - r )m1 + rm1 )P (1,1, B ) = lB P(0,0,V ) (5) ê 0 0 0 ú
- (lB +a1 + m1) rm1 a1
A1 =ê ú
+ b1P (3,1, B ) + m2 P (2,2, B) + (1 - r )m1P(1,2, B ) ê 0 0 0 - (lB +a2 + m2 ) 0 a2 ú
ê 0 0 0 0 ú
ê b 1 - (lB + b1) ú
- (l
ê 0 0 0 b 2
0 B + b )ú
(a1 + lB + (1 - r )m1 + rm1 )P(1, n, B) = lB P (1, n - 1, B ) (6) ë 2 û

+ b1 P(3, n, B) + m 2 P (2, n + 1, B ) + (1 - r )m1 P (1, n + 1, B), n > 1

émV 0 0 0 0 0ù
ê 0 0 0 0 0 0 ú
(lB + a2 + m2 )P(2,1, B) = rm1P(1,1, B) + b2 P(4,1, B) (7) ê ú
ê0 0 (1 - r )m 0 0 0ú
1
A2 = ê ú
ê 0 0 m2 0 0 0 ú
(lB + a 2 + m2 )P(2, n, B) = (8) ê 0 0 0 0 0 0 ú
rm1P(1, n, B) + b2 P(4, n, B) + lB P(2, n -1, B), n > 1 ê 0 0 0 0 0 0 ú
ëê ûú

Let X be the vector of the steady

state probabilities (l B + b1 ) P (3,1, B ) = a 1 P (1,1, B )
(9) with Q as coefficient matrix, such
that XQ=0, and the normalizing
condition is Xe=1, where e is the (lB + b1)P(3, n, B) = a1P(1, n, B) + lB P(3, n -1, B), n > 1 (l B + b 2 ) P ( 4,1, B )
column
(10) vector of appropriate dimension with all = a 2 P ( 2 ,1, B )
elements equal
(11) to 1. Let us partition X as X = [X 0 , X 1 , X
(lB + b2 )P(4,n, B) = a2P(2,n, B) + lB P(4, n -1, B), n > 1
X0
=[X ]
0 ,0 ,V
(12) X i = [ X 0 ,i ,V , X 1,i ,V , X 1, i , B , X 2 , i , B , X 3, i , D ,

We examine the existence of a solution of the form,

i-1 (13)
X i = X i -1 R or X i = X 1R ,i³1

Since, Xi depends only on the state transition between 3. MATRIX GEOMETRIC SOLUTION
level (i-1) and level i, the balance equations are given
by: The theory of matrix geometric approach was developed
X 0B00 + X1B10 = 0 (14) [26] to solve the stationary state probabilities for the
vector state Markov Process with repetitive structure.
X 0B01 + X1A1 + X 2 A2 = 0 (15)
Consider the generator matrix Q as shown below:
 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230 226

v Probability that the X i-1 A0 + X i A1 + X i+1 A2 = 0, i > 1 (16)

server is in broken down
state while failed In Equation (13), R is a square matrix and is the
during second optional unique minimal non negative solution to the non linear
service is: matrix equation
¥ (26) A0 + RA1 + R A2 = 0
2 (17)
P(D2 ) = å P(4,i, D )
i=1
The matrix R can be computed by successive
v Average number of customers in the system is: substitution in the recurrence relation R(0)=0
¥ 0 2 2
R ( n + 1) = - A
E[N]= å i [P (0 , i,V ) + P (1, i,V ) ]+ A -1 - R ( n) A A -1 , n ³ 0 (18)
1 1
i=0
¥ (27)
å i [P (1, i, B ) + P ( 2, i, B ) + P (3, i, D ) + P ( 4, i, D )]
i =1
Finally, we are interested to calculate the vector
v Throughput is given by:
¥ ¥
X=[X0, X1, X2,……] for this purpose. The balance
T (P) = mV å[P(0, i,V ) + P(1, i,V )] + å[m1P(1, i, B) + m2 P(2, i, B) (28) equations for the boundary states given by Equations
i =0 i =1 (14) and (15) can be written in matrix form as:
X (A1 + RA2 ) = 0 (19)
0

X (I - R)-1e = 1 (20)
5. SENSITIVITY ANALYSIS 0

where, e is the column matrix of suitable dimension

In order to determine the performance of the working having all elements 1; this gives a unique solution for
vacation queueing system with second optional service X=[X0, X1, X2,……].
and unreliable server, we perform a computational
experiment by a program developed in MATLAB using
matrix geometric technique discussed in section 3. The 4. PERFORMANCE MEASURES
Tables 1-5 and Figures 1-6 depict the variation of the
system performance characteristics with respect to the The validity of the model and the system performance
various input parameters. We assume the following characteristics can be analyzed by computing the system
basic input data for different tables: performance characteristics, in terms of the steady state
TABLE 1: λB=0.3, m1=6, m2=4, mV=2, probabilities explicitly. Some of the system performance
a1=0.05, a2=0.03, aV=0.02,
indices are as follows:
b1=3.5, b2=2.5, bV=1.5, r=0.2 v Probability that the server is on working vacation is
TABLE 2: λV=0.1, m1=6, m2=4, mV=2, obtained as:
¥
a1=0.05, a2=0.03, αV=0.02,
P (Wv ) = å P (0, i,V ) (21)
b1=3.5, b2=2.5, bV=1.5, r=0.2 i=0

TABLE 3: λV=0.1, λB=0.3, a1=0.05, a2=0.03,

v Probability that the server is in broken down state
αV=0.02, b1=3.5, b2=2.5, bV=1.5,
while fails during working vacation is:
r=0.2 ¥
TABLE 4: λV=0.1, λB=0.3, m1=6, m2=4, P (Dv ) = å P (1,i,V ) (22)
i =1
mV=2, b1=3.5, b2=2.5, bV=1.5
TABLE 5: λV=0.1, λB=0.3, a1=0.05, a2=0.03, v Probability that the server is busy with first
αV=0.02, r=0.2, m1=6, m2=4 essential service is:
¥ (23)
Tables 1 and 2 show the variation in long run P (B1 ) = å P (1, i, B )
probabilities of different states of the server by varying i =1
arrival rates. It is clear from tables that PWV and PDV
show decreasing trend with increased value of arrival v Probability that the server is busy with second
rates whereas PB1, PB2, PD1, PD2 have increased values optional service is:
¥ (24)
for higher arrival rates. Table 3 summarizes the effect of P (B2 ) = å P (2, i, B )
increased service rates on the long run probabilities. It is i =1
evident from table that the increased service rate results
v Probability that the server is in broken down state
in decreased PB1, PB2, PD1 , PD2, keeping the working
while failed during first essential service is:
vacation service rate µV as constant. Table 4 illustrates ¥
P(D1 )= å P(3, i, D) (25)
that the probabilities of the server being busy and under
i=1
227 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230

average queue length but later on a sharp decrement can repair state, follow increasing trend for increased server
be seen in E[N] on increasing the repair rate of the breakdown rates α1, α2 whereas PWV and PDV have very
server. Therefore, by keeping higher repair rate, the dormant decrement with the increase in failure rates.
performance of the system can be improved. Figures 4-6 Table 5 presents the comparison between probabilities
depict how service rates affect the average queue length. of server status for different repair rates β1 and β2 by
keeping βV as constant. We observe that the vacation
TABLE 3. Long run probabilities of the server status by state probabilities of the server increase with the
increase in the value of repair rate while the
varying µ1 and µ2
probabilities of the server being busy and under repair
µv µ1 Pwv PDv PB1 PB2 PD1 PD2
show decreasing pattern.
2 0.0054 0.0017 0.3369 0.3169 0.2718 0.1720 In Figures 1-6, we plot the variation of the queue
4 0.2329 0.1104 0.2912 0.2778 0.1841 0.0882 length with respect to different parameters such as
6 0.2607 0.2199 0.1678 0.2542 0.1477 0.0795 failure rate, repair rate, service rate, etc.. Figures 1-6
8 0.2890 0.2634 0.1082 0.2091 0.1191 0.0651 depict that the average queue length E[N] follow an
increasing trend for higher values of arrival rate λ.
3 µ2 Pwv PDv PB1 PB2 PD1 PD2 In Figures 1 and 2, E[N] depicts a clearly increasing
1 0.0082 0.0024 0.3301 0.3028 0.2601 0.1102 pattern with the increase in server breakdown rate. This
3 0.2408 0.1212 0.2721 0.2493 0.1792 0.0828 trend indicates that the system performance
5 0.2841 0.2652 0.1390 0.2391 0.1408 0.0729 tremendously affected by the server breakdown and will
lead to increased congestion. From Figures 3 and 4, it is
7 0.2988 0.2981 0.0998 0.1830 0.1118 0.0539
noticed that initially there is a gradual decrement in the

TABLE 4. Long run probabilities of the server status by TABLE 1. Long run probabilities of the server status by
varying α1 and α2 varying λv
αv α1 Pwv PDv PB1 PB2 PD1 PD2 λv Pwv PDv PB1 PB2 PD1 PD2
0.01 0.0078 0.0025 0.5328 0.3194 0.1691 0.0174 0.1 0.4071 0.3029 0.1040 0.1685 0.1147 0.0785
0.03 0.0064 0.0020 0.6210 0.3209 0.1729 0.0205 0.2 0.4009 0.3015 0.1301 0.1725 0.1194 0.0975
0.05 0.0052 0.0016 0.6350 0.3231 0.1832 0.0247 0.3 0.3498 0.2222 0.1506 0.2256 0.1393 0.1084
0.07 0.0043 0.0011 0.6520 0.3301 0.1929 0.0391 0.4 0.3050 0.2107 0.1810 0.2520 0.1577 0.1137
0.02 Pwv PDv PB1 PB2 PD1 PD2 0.5 0.3005 0.2072 0.1978 0.2612 0.1654 0.1262
α2
0.02 0.0059 0.0019 0.5197 0.3273 0.1337 0.0431 0.6 0.2777 0.2047 0.2437 0.2734 0.1676 0.1373
0.04 0.0056 0.0014 0.5431 0.3309 0.1568 0.0910
0.7 0.2119 0.1453 0.2556 0.3661 0.1666 0.1404
0.06 0.0052 0.0011 0.5578 0.3746 0.1691 0.0943
0.8 0.1439 0.1317 0.3271 0.5934 0.1635 0.1469
0.08 0.0049 0.0009 0.5629 0.3809 0.1731 0.0971
0.9 0.0994 0.0932 0.3619 0.5812 0.1585 0.1557

TABLE 5. Long run probabilities of the server status by TABLE 2. Long run probabilities of the server status by
varying β1 and β2 varying λB
βv β1 Pwv PDv PB1 PB2 PD1 PD2 λB Pwv PDv PB1 PB2 PD1 PD2
3 0.0951 0.0216 0.5374 0.3294 0.1148 0.1481 0.1 0.5975 0.4833 0.0016 0.0519 0.0172 0.0233
5 0.0953 0.0217 0.4438 0.3019 0.1031 0.1423 0.2 0.4820 0.4847 0.0034 0.0819 0.0212 0.0476
7 0.0954 0.0217 0.3368 0.3000 0.1009 0.1409 0.3 0.4660 0.4799 0.0055 0.1191 0.0264 0.0743
9 0.0959 0.0319 0.2291 0.2902 0.0962 0.1399 0.4 0.3493 0.4749 0.0080 0.1641 0.0322 0.1054
2.5 Pwv PDv PB1 PB2 PD1 PD2 0.5 0.3322 0.3697 0.0111 0.2179 0.0381 0.1434
β2
2 0.0944 0.0314 0.5242 0.3616 0.1443 0.1988 0.6 0.3147 0.3645 0.0150 0.2819 0.0437 0.1917
4 0.0958 0.0316 0.5297 0.3220 0.1201 0.1701 0.7 0.2968 0.2591 0.0202 0.3582 0.0486 0.2546
6 0.0961 0.0319 0.4038 0.3201 0.1179 0.1516 0.8 0.1780 0.1536 0.0269 0.4500 0.0524 0.3385
8 0.0979 0.0324 0.3981 0.2201 0.2124 0.0998 0.9 0.1600 0.0480 0.0358 0.5630 0.0551 0.3520
 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230 228

9
8
7 m1=m2=3
8 a1=a2=0.3
7 a1=a2=0.9
6
m1=m2=5 6 a1=0.2,a2=0.4
m1=3,m2=7
E[N]

5 a1=0.95,a2=0.85

E[N]
5
4 m1=4,m2=2 4
3 3
2 2
1 1
0 0
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5
l l
Figure 5. Average queue length vs l Figure 1. Average queue length vs l
10

10 av =a1=a2=0.2
9 mv=m1=m2=2
8 8 av =a1=a2=0.7
7 mv=m1=m2=4 6 av =0.15,a1=0.2,a2=0.25
mv=1,m1=3,m2=2
6 E[N] av =0.5,a1=0.9,a2=0.7
E[N]

mv=3,m1=5,m2=4
5 4
4
3 2
2
1 0
0
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5
l
l
Figure 6. Average queue length vs l Figure 2. Average queue length vs l
23

6. CONCLUDING REMARKS 21 b1=b2=2

19
b1=b2=5
In many stochastic systems, there may occur a 17 b1=2.5,b2=1.5
E[N]

situation in which the first service is essential to all 15 b1=3.5,b2=6.5

13
arrivals whereas second service is needed by only
11
some of them. Similarly breakdown is a remarkable 9
and unavoidable phenomenon in the service facility 7
of a queueing system, because system performance 5
deteriorates seriously by the server breakdown and 0.1 0.2 0.3 0.4 0.5
limitations of the repair capacity. In this paper, we l
have studied the effect of second optional service and
Figure 3. Average queue length vs l
unreliable server on the performance measures for 23
working vacation queueing model. The queue with
working vacation may be applicable in modeling of
21 bv =b1=b2=2
many practical situations related to computers,
communications, and productions systems, etc., 19 bv =b1=b2=4

wherein the server works at different service rates 17 bv =1,b1=3,b2=2 bv

=3,b1=5,b2=4
rather than completely stopping the service during a
E[N]

15
vacation. 13
The inclusions of realistic factors such as 11
9
unreliable server, optional service, working vacation,
7
etc. make our model more versatile from application
5
point of view. Using matrix geometric solution we
0.1 0.2 0.3 0.4 0.5
have obtained some important performance
measures, which may be useful for the system l

designers and practitioners involved in many Figure 4. Average queue length vs l

industrial organizations operating in congestion
scenario.
229 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230

16. Ke, J.C., “Batch arrival queue under vacation policy with server 7. REFERENCE
breakdown and startup/closedown time”, Applied Mathematical
Modelling, Vol. 31, (2007), 1282-1292.
17. Ke, J.C., “The optimal control of an M/G/1 queueing system 1. Takagi, H., “Queueing Analysis Vacation and Priority Systems”,
with server vacations, startup and breakdowns”, Computers & Vol. 1, North Holland, Amsterdam, (1991).
Industrial Engineering, Vol. 44, (2003), 567-579.
18. Wang, K.H., Wang, T.Y. and Pearn, W.L., “Optimal control of 2. Lee, S.S., Lee, H.W., Yoon, S.H. and Chae, K.C., “Batch arrival
the N-policy M/G/1 queueing system with server breakdowns queue with N-policy and single vacation”, Computers &
and general startup times”, Applied Mathematical Modelling, Operations Research, Vol. 22, (1995), 173-189.
Vol. 31, (2007), 2199-2212. 3. Chao, X. and Zhao, Y.Q., “Analysis of multiserver queue with
19. Yang, X. and Alfa, A.S., “A class of multi server queueing station and server vacation”, European Journal of Operational
system with server failure”, Computers & Industrial Research,Vol. 110, (1998), 392-406.
Engineering, Vol 56 (1), (2009), 33-43. 4. Bacot, J.B. and Dshalalow, J.H., “A bulk input queueing system
20. Madan, K.C., “An M/G/1 queueing system with second optional with batch gated service and multiple vacation policy”,
service”, Queueing System, Vol. 34, (2000), 37-16. Mathematical and Computer Modelling, Vol. 34, (2001), 873-
21. Medhi, J., “A single server Poisson input queue with a second 886.
optional channel”, Queueing Systems, Vol. 42, (2002), 239-242. 5. Seri, L.D. and Finn, S.G., “M/M/1 queues with working
22. Al-Jararha, J. and Madan, K.C., “An M/G/1 queue with second
vacations (M/M/1/WV)”, Performance Evaluation, Vol. 50,
optional service with general service time distribution”,
(2002), 41-52.
Information Management Science, Vol. 14, (2003), 17-56.
6. Arumuganathan R. and Jeyakumar S., “Steady state analysis of
23. Choudhury, G. and Paul, M., “A batch arrival queue with an
additional service channel under N-policy”, Applied bulk queue with multiple vacations, setup times with N-Policy
Mathematical Modelling, Vol. 156, (2004), 115-130. and closedown times”, Applied Mathematical Modelling, Vol.
24. Wang, J., “An M/G/1 queue with second optional service and 29, (2005), 972-986.
server breakdowns”, Computers and Mathematics with 7. Wu, and Takagi, “M/G/1 queue with multiple working
Applications, Vol. 47, (2004), 1713-1723. vacations”, Performance Evaluation, Vol. 63, (2006), 654-681.
25. Choudhury, G., Ke, J.C. and Tadj, L., “The N-policy for an 8. Tian, N. and Zhang, Z.G., “A two-threshold vacation policy in
unreliable server with delaying repair and two phase of service”, multiserver queueing system”, European Journal of Operational
Journal of Computational and Applied Mathematics, Vol. 231, Research, Vol. 168, (2006), 153-163.
No. 1, (2009), 349- 364.
9. Ke, J.C., “Operating characteristics analysis of Mx/G/1 system
26. Neuts, M.F., “Matrix Geometric Solution for Stochastic Models”
Johns Hopkins University Press, Baltimore, MD, (1981). with variant vacation policy and balking”, Applied
Mathematical Modelling, Vol. 31, (2007), 1321-1337.
27. Gomez, C.A., “A matrix geometric approximation for tendem
queues with blocking and repeated attempts”, Operations 10. Cao, J. and Cheng, K., “Analysis of M/G/1 queueing system
Research Letters, Vol. 30, (2002), 95-128. with repairable service station”, Acta Mathematicae Applicatae,
28. Choi, S.H., Kim, B. and Soharaby, K., “On matrix geomertric Vol. 5, (1982), 113-127.
solution of nested QBD chains”, Queueing Systems, Vol. 43, 11. Li, W., Shi, D. and Chao, X., “Reliability analysis of M/G/1
(2003), 95-128. queueing system with server breakdowns and vacations”,
29. Gray, W.J., Patrick, P. and Scott, M., “ A queueing model with Journal of Applied Probability, Vol. 34, (1997), 546-555.
multiple types of server breakdowns”, Quality Technology and 12. Wang, J., Cao, J. and Li, Q., “Reliability analysis of the retrial
Quantitative Management, Vol. 1, (2004), 245-255.
queue with server breakdown and repairs”, Queueing Systems,
30. Ke J. C. and Wang K. H., “Vacation policies for machine repair Vol. 38, (2001), 363-380.
problem with two types of spare”, Applied Mathematical
Modelling, Vol. 31, (2007), 880-884. 13. Wang, K.H., Chan, M.C. and Chuan, J., “Maximum entropy
31. Jain, M. and Jain, A., “Working vacation queueing model analysis of Mx/M/1 queueing system with multiple vacations
multiple types of server breakdown”, Applied Mathematical and server breakdowns”, Computers & Industrial Engineering,
Modelling, Vol. 34, No. 1, (2010), 1-13. Vol. 52, (2007), 192-202.
14. Almasi, B., Roszik, J. and Sztrik, J., “Homogeneous finite source
retrial queues with server subject to breakdowns and repair”,
Mathematical and Computer Modelling, Vol. 42, (2005), 673-
682.
15. Sherman, N.P. and Kharoufeh, J.P., “An M/M/1 retrail queue
with unreliable server”, Operations Research Letters, Vol. 34,
(2006), 697-705.
 M. Jain and D. Chauhan / IJE TRANSACTIONS C: Aspects Vol. 25, No. 3, (September 2012) 223-230 032

Working TECHNICAL NOTE

Vacation
Queue
with
Second
Optional
Service
and
Unreliable
Server

M. Jain a, D . Chauhan b

a Department of
Mathematics, Indian
Institute of Technology,
Roorkee-247667
b Department of
Mathematics,
Allenhouse Institute
of Technology,
Kanpur-208007
:Keywords INFO ARTICLE
Working Vacation
Second Optional Service
Matrix Geometric Technique
Server Breakdown :Article history
Repair Received 2 January 2010
Queue Size Accepted in revised form 17 May 2012

doi: 10.5829/idosi.ije.2012.25.03c.06