• Aucun résultat trouvé

On the busy period for the M|G|1 queue with-finite waiting room

N/A
N/A
Protected

Academic year: 2022

Partager "On the busy period for the M|G|1 queue with-finite waiting room"

Copied!
5
0
0

Texte intégral

(1)

R

EVUE FRANÇAISE D

AUTOMATIQUE

,

INFORMATIQUE

,

RECHERCHE OPÉRATIONNELLE

. R

ECHERCHE OPÉRATIONNELLE

E. J. V ANDERPERRE

On the busy period for the M|G|1 queue with-finite waiting room

Revue française d’automatique, informatique, recherche opé- rationnelle. Recherche opérationnelle, tome 8, noV1 (1974), p. 41-44

<http://www.numdam.org/item?id=RO_1974__8_1_41_0>

© AFCET, 1974, tous droits réservés.

L’accès aux archives de la revue « Revue française d’automatique, in- formatique, recherche opérationnelle. Recherche opérationnelle » implique l’accord avec les conditions générales d’utilisation (http://www.numdam.org/

conditions). Toute utilisation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fi- chier doit contenir la présente mention de copyright.

Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques

http://www.numdam.org/

(2)

ON THE BUSY PERIOD FOR THE M|G|1 QUEUE WITH FINITE WAITING ROOM

by E. J, VÂNDERPERRE (*)

Summary. — The Laplace-Stieltjes transform (L.5.T.) of the probability distribution function of the busy period for the M\G 1 queueing System with finite waiting room can be

obtained by using a supplementary variable method (a Markov characterization).

In the present paper > it is shown that applying this method, it is rather easy to analyse queueing Systems with a finite waiting room and with Poisson input.

1. INTRODUCTION

For the well known model of the M|G|l queue with finite waiting room, Cohen [1] obtained the L.S.T. of the distribution function of the busy period in a rather simple way using a Markov renewal branehing argument.

In this paper, we use a simple method based on the inclusion of a supple- mentary variable (Markov characterization). Moreover, the method is not restricted to queueing Systems with Poisson input.

2. MATHEMATTCAL ANALYSES OF THE SYSTEM

Dénote by X"*1 the mean interarrivai time of customers and by F(.) the distribution function of the service times. Lét QK dénote the duration of a busy period of the M|G|l queue with K waiting places. Acustomer who finds ail waiting places occupied, cannot enter the system. He départs and never returns (overflow).

Moreover, let Xt be the number of customers waiting in the system at any instant of time t and xt the past service time of the customer being served at time t. It is assumed that a customer arrives at time / = 0 and that he meets an empty system.

(1) Département de sciences économiques, Université de Hasselt, Belgique.

Revue Française d^Automatique, Informatique et Recherché Opérationnelle n° jan. 1974, V-l.

(3)

42 E. J. VANDERPERRE

The busy period 0* is the time between t = 0 and the first instant at which the system becomes empty (the absorbing state).

For i = 0, 1,..., K and /, x > 0 we define

x)dx M Pr[Xt = i,x^xt<x + dx\xo = 0]

By simple probabilistic arguments, we have for i = 0, ls . . . K— 1

- x, 0 - F(x)}

P& x) - £" Pj(t - x, 0 +) { 1 - E e-*" ^ } [1 - F(*)]

The Laplace transform with respect to t of pt (f, x) will be denoted by pï(s9 x) ; Re s > 0.

If we consider the events in which xt = 0 +5 then we obtain the following set of récurrence relations for the boundary values of the functions pf (s, x) ;

(i)

£

o+)= 2- Pîfco

where Sof is the Kronecker symbol.

Let

/•oc

m * Jo

ee-x*

i+l-J

Res ^ 0

i> 0

The L.S.T. of the probability distribution function of the busy period 6K will be denoted by E { c~s^ }.

Clearly

(2) E {*-**}= Ms)pS(s9O+)

Revue Française d'Automatique, Informatique et Recherche Opérationnelle

(4)

The foUowing column K-vectors will be used.

f = [1,0, ...,0, ...,0]T

and the K x K matrix function M>) Ms) 0 0 Ms) Ms) Ms) 0

Ms) M') À® Ms

P(s) =

o

fi+1(s) ƒ,(*) ƒ,_!(*) . . . M*) Ms)

where

By (1) we obtain

0 0 0

0 0 0

0 0 0

oo

ao(s)

where / is the unit matrix.

It can be shown that the'matrix [ 7 ~ P ( r ) ]90 < Re 5 < 00, is nonsingular.

For AT > 0 we define det [7—P(s)] = AK(^) and A0(*y) = 1. By (2) and Cramer5s rule, we have

(3)

But (4)

Let (5)

E { e" & } = I.

;~ '

n» janvier 1974, V-l.

(5)

4 4 E. J. VANDERPERRE

By (4) and (5) we obtain after some simplifications

(6) YK(S) =J E * fj(s)YK+i-j(s) + aK.t(s) K>-\

Let

CO Y ( * Z ) « * S YK(S)ZK \Z\ < \9(s)\

where ç>(s) is the smallest root of the f unctional équation Z —f(s + X —XZ) = 0 Re s ^ 0.

By (6), (7) and algebra, we find

If O dénote a circle with center at the origin of the complex Z-plane and with radius \z\ < |pO)|, Re s > 0, then by (3), (5) and (8) we obtain for K= 1,2, . . . R e j ^ 0,

ï - {ï -M } ö^f ^ ï ( i - zy'iz-fis + E {e-4* } = füJ?

i — {î —fis) }^^=n — zf

1

[z—fis + x — xz)]"

1

It is easy to show that the result agrées with Cohen's Theorem [1] p. 825.

REFERENCES

[1] COHEN J. W., On the busy periods for the M\G\1 queue withfinite and with infinité waiting room, J. Appl. Prob., 8 (1971), 821-827.

[2] VANDERPERRE E. J., The[busy period ofa repairman attending a n + 1 unit parallel System, R.AJ.R.O, V-2 (1973), 124-126.

Revue Française d'Automatique, Informatique et Recherche Opérationnelle n° jan. 1974, V-l.

Références

Documents relatifs

This is the baseline survey of a prospective cohort study of adult patients selected during a 1-week period from the waiting rooms of 12 primary care clinics within the

I had just witnessed a small monumental waiting room moment—the unintended consequences of having magazines in waiting rooms.. My wife and I visited the Mayo Clinic in Rochester,

Restructuring the waiting room and its role in health care is a potential way to improve patient and doctor satisfaction, as well as consultation efficiency.. Using the waiting

In the perturbation theory of the Boltzmann equation for the global well-posedness of solutions around Maxwelians, the energy method was first de- veloped independently in [16, 14]

The winding number of a closed curve around a given point is an integer representing the total number of times that curve travels anti-clockwise around the point.. The sign of

(b) The previous question shows that it is enough to consider the case k, l 6 n/2, for otherwise we can replace the representations by isomorphic ones so that these inequalities

Keywords : queueing theory, regular variation, supremum of actual, virtual waiting time, limit theorems, extreme value

When visiting the emergency department, the perception of the time spent in the waiting room before the beginning of the care, may influence patients' experience.. Based on models