Escolar Documentos
Profissional Documentos
Cultura Documentos
Abstract— Here arrivals of the customer follow poisson distribution and there is single server who is providing service to the
customer. Also server serve the customer on first come first serve basis. The vacation queues have been extended to computer networks,
communications systems, as well as production management, inventory management and other fields (Doshi B.T., 1990). For the classical
single-server vacation model, a server stops working completely during the vacation periods. However, the server can be generalized to work at
a different rate during the vacation periods. This type of queueing models are useful to analyze a variety of multi-queue systems that frequently
arise in computer and communication networks. Different parameter is used for finding expected busy period and queueing length.
Keywords— Arrival pattern is poisson, Removable server, Service time is exponential distribution, queue discipline is FCFS, Stochastic process.
__________________________________________________*****_________________________________________________
4) Server serve the customer single if batch size is less
I. INTRODUCTION
than „a‟
In this paper, the arrivals process is assumed to be Poisson 5) The Queue discipline is first come first serve
with parameter and the service by a single server. If the 6) The various stochastic process in the system are
queue is empty then server goes on vacation, when the statistically independent
queue size is less then „a‟ the server is on working vacation 7) The server is removed from its service as soon as the
& if the queue size is > „a‟ then server is busy. The server Queue is empty.
serve the customer according to exponential law with
parameter 1 if server is on working vacation & the server II. Analysis of Model
serve the customer according to exponential law with
Pi,n(t) = P{x(t) = i, y(t) = n}
parameter 2 if server is busy. For example, the allocation of
the real time of a processor within a large switch or the x(t) = 0 y(t) = 0 vacation
allocation of the bandwidth of the bus of a LAN. In such x(t) < a y(t) = 1 working vacation
systems, each queue may be processed either at a fast
service rate or a nominal service rate depending on whether x(t) > a y(t) = 2 busy
it acquires a token or not: That is, a queue operates at its fast The difference-differential equation governing the
service rate as soon as it gets the token. The scheduling
model are
discipline is exhaustive. In an exhaustive schedule, a queue
possesses the token until the queue is empty. And then, the ( t ) = P00(t) + 1 P0,1(t) + 2 P0,2(t)
P00 (1)
token is passed to the next queue. From the point of view of
the individual queue, the server follows a working vacation. P0,1 (t ) (λ μ)P0,1 (t ) λP00 (t ) μ1P11 (t ) μ 2 P12 (t )
The server works at a slower rate rather than completely
stops during a vacation. Such type of situation are called
working vacation. Liu, Xu and Tian (2002), Takagi (2006), (2) Pn ,1 (t )) (λ μ1 )Pn ,1 ( t ) + Pn1(t)
W.D. (2006), Servi, L.D. (2002) are discussed some
working vacation queue. + 1Pn+1,11(t) 1na1
(3)
Assumptions: The following assumption describe the
system P0, 2 ( t ) = ( + 2) P0,2(t) + Pa1,1(t) + 2 Pn,2(t)
1) Arrivals arrive according to poisson law with
parameter n>a (4)
2) The service time distribution is exponentially with rate Pn , 2 = ( + 2) Pn,2(t) + Pn1,2(t) + 2 Pn+a,2(t)
1 and 2
3) Server serve the customer of batch sized if server is n1 (5)
busy
163
IJRITCC | January 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 1 163 – 166
_______________________________________________________________________________________________
Taking Laplace Transformation of equation (1) to (5) (S+ + 2) Pn , 2 (s) λPn 1, 2 (s) μ 2 Pn a , 2 (s)
SP00 (s) P00 (0) P0,0 (s) +1 P0,1 (s) μ 2 P0, 2 (s) h(z) = za+1 (S + + 2)z +
(S + ) P00 (s) 1 μ1P01 (s) μ 2 P0,2 (s) P00(0) = 1 h(z) = 0
(S + + ) P01 (s) λ P00 (s) μ1P11 (s) μ 2 P12 (s) Consider the circle |z| = 1 where is arbitrarily
Pn1 (s) λPn1,1 (s) μ1Pn1,1 (s) μ 2 Pn1,2 (s) (8) (z) = 1.
Since f(z) has only one zero inside the circle
1
Pn ,1 (s) (λPn 1,1 (s) μ1 Pn 1,1 (s) μ 2 Pn 1, 2 (s) f(z) + g(z) = h(z) will also have only one zero
S λ μ
inside f(z) = 1, this root of h(z) = 0 is real and
P0,2 (s) P0,2 (0) (λ μ 2 ) P0,2 (s) λPa 1,1 (s) μ 2 Pn ,2 (s)) unique
(S λ μ 2 ) P0, 2 (s) λPa 1,1 (s) μ 2 Pn ,2 (s) λ
iff = 1 & (0 < < 1)
aμ1
1
P0, 2 (s) (λPa 1,1 (s) μ 2 Pn , 2 (s)) and other roots i 1
(S λ μ 2 )
Then satisfies the equation
Pn ,2 (s) Pn 2 (0) (λ μ 2 ) Pn, 2 (s) λPn1,2 (s) μ 2 Pna ,2 (s) λ α(1 α a )
(9) a = = + q +…+ a
μ1 1 α
λ α(1 α a )
(S++2) Pn ,2 (s) λPn1,2 (s) μ 2 Pna , 2 (s) and
μ2 1 α
(10)
From equation (11)
The characteristic equation from equation (10)
(S λ μ ) (S λ μ1 ) 2 4μ 2 λ
(S ++1) Pn1 (s) λPn1,1 (s) μPn1,2 (s) z=
2μ 2
h(z) = 2 z (S+ + 1) z +
2
Let
h(z) = 0
(S λ μ ) (S λ μ1 ) 2 4μ 2
2 z2 (S + + 1)z + = 0 =
2μ 2
(11)
From (10)
164
IJRITCC | January 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169
Volume: 6 Issue: 1 163 – 166
_______________________________________________________________________________________________
(S λ μ ) (S λ μ1 ) 2 4μ 2 P00 = P μ1 μ 2 K
01
= λ λ
2μ 2
λ2 α
K=
Let is the unique positive real root λ μ )(λ μ ) μ λ2 αR n
1 2 2
Hence by Rouche‟s theorem
n
Pn ,1 (s) P0,1 (s)α P0,2 (s)μ 2 R n Pn1 = P01 λ Kμ R n
μ1 2
μ1 μ 1
P00 (s) P01 (s) 2 P01 (s).D1
S λ S λ S λ L qB = n Pn 2
n a
P01 (s) 1
P00 (s) (μ1 μ 2 D1 )
2 R n
S λ S λ L qB = P01 n ( )( ) R
na
2 n
1 2 2
n 1
Pn ,1 (s) P01 (s)[α Dμ 2 R ]
Expected Busy Periods
n 2
R λα
Pn , 2 (s) P01 (s) E(idle Period )
(S λ μ1 )(S λ μ 2 ) μ 2 λ2 αR n P00 =
E(idle Period ) E(Busy Period )
According to normalizing condition
n
1
1 E(IdP) =
Pi0 (s) Pi,1 (s) Pi2 (s) S λ
i 0
1
Steady state probabilities P00 = λ 1
P00
Pin = Lim S Pi,n (s)
1
E(BP ) 1 λE(BP )
s λ
References
0.06 0.2 0.3 0.16666667 0.23310346 [1] Ashok Kumar , “Application of Markovian process to
queueing with cost Function PhD thesis, KU, Kurukshetra ”
0.06 0.3 0.3 0.25 0.89839085 1979
[2] Takagi H, “Mean message wasting time in asymmetric
polling system Elsevier science publishers BV (North
Holland)1990.
[3] Chaudhary M.L and G C O “The queuing system M/GB/1
and its ramifications operation research-6” pp. 56-60.
[4] Ahmed M.M.S “Multi-channel bi-level heterogeneous
servers, bulk arrivals queueing system with Erlangian
service time”. Mathematical & computational applications,
vol. 12 pp. 97-105. 2007
[5] Tuteja R.K “Some studies in the theory of queues with
special reference to correlated queueing “PhD thesis KUK
(1971).
[6] Vinod B, “Exponential queries with server vacation, J
.operation research sco. Vol. 37 pp 1007- 1014.
Table - II
P01 2 A Lqwv
166
IJRITCC | January 2018, Available @ http://www.ijritcc.org
_______________________________________________________________________________________