STOCHASTIC DECOMPOSITION PROPERTY
For the system under consideration, we establish the decomposition of the random variable representing the system size in two different ways.
- 1. The number of customers L, in the system can be written as L = L' + M', where L' is the number of customers in “the standard Geo IG11 queue with impatient customers and starting failures” and M' is the number of retrial customers when the server is free.
- 2. The system size L can also be rewritten as L = L" + M", where L" is the system size in “the Geo / G /1 queue with impatient customers” and M" is the number of customers retrying for service when the server is free or under repair.
Proof. We can easily rewrite equation (6.32) in the following two different ways.
Here, the first term is the PGF of the system size in “Geo / G /1 queue with starting failure” and the second term of the product is the PGF of the number of retrial customers when the server is free.
In this representation of
Geo IG11 standard queue and the other term is the PGF of the number of customers retrying their service when the server is idle or under repair.
Exhaustive numerical results are presented in this section to describe the behavior of the queueing systems under study. We examine the effect of the system parameters such as arrival probability (a), balking probability (b). feedback probability (v) and the probability that the server is started (0) to serve the customer successfully on some of the crucial performance measures of our model. The values of the parameters are chosen appropriately and the numerical results are obtained by using MATLAB software.
Figure 6.1 (a)-(l) depicts the impact of a,b,% and v on P(0). As expected intuitively, P(0), the probability that the system is empty, decreases with the increase in a for any values of b.d and v (Figure 6.1(a)-(c)). Further, P(0) is almost stable for increasing values of b. We can see from Figure 6.1 (d)-(f), P(0) remains unaltered as for as b is concerned but it is higher for higher values of a,Q and also v. Figure 6.1 (g)-(i) shows an increasing trend for P(0) for increasing values of 0. Figure 6.1 (h) confirms the observation that P(0) is not much affected by b. P(0) is plotted against v for different values of a,b and 0 in Figure 6.1 (j)-(l). We can see from these Figure 6.1 that P(0) decreases with the increment in v. Further, we can see it increases with increase in b and 0 but decreases with increase in a.
The effect of a,b,% and v on E(L) is plotted in Figure 6.2(a)-(l). The behavior of E(L) with respect to a is described in Figure 6.2(a)-(c). It is noticed that the
FIGURE 6.1 Impact of and v on P(0).
FIGURE 6.1 (Continued)
expected system size E(L) increases for increasing values of a which matches with our intuition. Figure 6.2(d)-(f) showcases the trend of E(L) for increasing values of b. E(L) decreases as b increases. Also it is observed that E(L) is higher for higher values of a and v but it is the other way for 0. E(L) is drawn for increasing values of 0 for different values of a,b and v in Figure 6.2(g)-(i). As is expected, E(L) decreases for increasing values of 0 in all the three different cases. Also, E(L) is higher for the higher values of a and v but is very low for higher values of b. Finally, in Figure 6.2(j)-(l), E(L) is depicted for increasing values of v for various values of a,b and 0. We observe in Figure 6.2 an increasing trend for E(L) for increasing v for a,b and 0. Further, E(L) is higher for the higher values of a but it is on the other way for b and 0. Also, we see a different pattern when a is very small (a = 0.05) and b is more (b = 0.6) which may be the combined effect of v and a (Figure 6.2(j)) and v and b (Figure 6.2(k)). The numerical results established above demonstrates that our analytical results are valid. Further, they can provide some information required for the concerned system designers for making optimal designs.
In the present study, we considered a Geo IG /1 queue with retrial customers. We assume that the server is prone to starting failure and the customer may rejoin the system for another service if his service is not satisfactory. Further, we have taken into consideration, the customer’s dilemma whether to join the system or leave if he or she anticipates a longer waiting time. We have analyzed the system in the early arrival set up where the departure is having precedence over the arrivals. It
FIGURE 6.2 Impact of a, 6.0 and V on E(L).
FIGURE 6.2 (Continued)
is observed that the P(0) increases for increasing values of 0, the probability that the server is started successfully and decreases with increasing value of feedback probability v. Further, it is noticed that the expected system size increases with the increase in the value of v and decreases with the increase in the values of 0. The trends shown in the graphs are in line with our intuition, which ensures the correctness of the analytical results obtained. Our model can be used to study a flexible manufacturing system where the arrival of components and their service time are occurring at regularly spaced time epochs. The arrival of the components, the manufacturing unit, a pool of waiting components, the failure of the system, unsatisfied components processed again may correspond to the customers, the server, retrial customers, starting failure and feedback, respectively. The future work may include study of the same system in late arrival setup, waiting time distribution and cost analysis considering holding cost and the cost of losing a customer.
The authors thank the referees for their valuable input in improving the manuscript.
Aboul-Hassan A., Rabia S. I., and Kadry A. 2005. Analytical study of a discrete time retrial queue with balking customers and early arrival scheme. Alexandria Engineering Journal 44:919-925.
Aboul-Hassan A., Rabia S. I., and Kadry A. 2008. A discrete time Geom/G/1 retrial queue with general retrial times and balking customers. Journal of the Korean Statistical Society 37:335-348.
Artalejo J. R. 1999. A classified bibliography of research on retrial queues: progress in 1990— 1999. Гор 7:187-211.
Artalejo, J. R. 2010. Accessible bibliography on retrial queues: progress in 2000-2009. Mathematical and Computer Modelling 51:1071-1081.
Artalejo J. R., and Gomez-Corral A. 2008. Retrial Queueing Systems: A Computational Approach, Springer, Berlin.
Atencia I., Ruiz I. F. and Sanchez S. 2009. A discrete-time retrial queueing system with starting failures, Bernoulli feedback and general retrial times. Computers & Industrial Engineering 57(4): 1291-1299.
Atencia, I. 2017. A Geo/G/1 retrial queueing system with priority services, European Journal of Operational Research. 256:178-186.
Atencia I., and Moreno P. 2004a. A discrete-time Geo/G/I retrial queue with general retrial times. Queueing Systems 48:5-21.
Atencia I., and Moreno P. 2004b. A discrete-time Geo/G/I retrial queue with optional service. Proceedings of the Fifth International Workshop on Retrial Queues 9:117-121.
Atencia I., and Moreno P. 2006. A discrete-time Geo/G/I retrial queue with the server subject to starting failures. Annals of Operational Research. 141:85-107.
Bruneel, H. and Kim B.G. 1993. Discrete-Time Models forCommunication Systems Including ATM. Kluwer Academic Publishers, Boston.
Choi B. D., and Kim J. W. 1997. Discrete-time Geol,Geo2/G/l retrial queueing system with two types of calls. Computers and Mathematics with Applications 33:79-88.
Gao S., and Wang X. 2019. Analysis of a single server retrial queue with server vacation and two waiting buffers based on ATM networks. Mathematical Problems in Engineering 1-14. https7/doi.org/10.1155/2019/4193404.
Gomez-Corral A. 2006. A bibliographical guide to the analysis of the retrial queues through matrix analytic techniques. Annals of Operations Research 141:163-191.
Krishna Kumar B., Pavai Madheswari S. and Vijayakumar A. 2002. The M/G/I rertial queue with feedback and starting failures. Applied Mathematical Modelling 26:1057-1075.
Kulshrestha, R. and Shruti 2020. Discrete-Time Analysis of Communication Networks with Second Optional Service and Negative User Arrivals. In Lecture Notes in Networks and Systems. Springer during ICCIS 2019 at SKIT. Jaipur, India.
Lan S., and Tang Y. 2019. Performance analysis of a discrete-time Geom/G/1 retrial queue with non-preemptive priority, working vacations and vacation interruption. Journal of Industrial & Management Optimization 15:1421-1446.
Madan, K.C. and AI-Rawwash, M. 2005. On the M X /G/l queue with feedback and optional server vacations based on a single vacation policy. Applied Mathematics and Computation. 160(3): 909-919.
Meisling T. 1958. Discrete time queueing theory. Operations Research 6:96-105.
Nobel R. 2016. A mixed discrete time delay/retrial queueing model for handover calls competing for a target channel. Queueing Theory and Network Applications 383:173-185.
Pavai Madheswari S., Krishna Kumar B., and Suganthi P. 2019. Analysis of M/G/I retrial queues with second optional service and customer balking under two types of Bernoulli vacation schedule. RAIRO-Operational Research 53:415-443.
Takacs L. 1963. The Stochastic law of the busy period for a single server queue with poisson input. Journal of Mathematical Analysis and applications 6:33-42.
Takagi H. 1993. Queueing Analysis: A Foundation of Performance Evaluation, Discrete- Time Systems. North-Holland, Amsterdam.
Upadhyaya S. 2018. Performance analysis of a discrete-time Geom/G/1 retrial queue under J-vacation policy. International Journal of Industrial and Systems Engineering 29:369-388.
Wang J., and Zhao Q. 2007. A discrete time Geom/G/1 retrial queue with starting failures and second optional service. Computers & Mathematics with Applications 53:115-127.
Wei C., Cai L. and Wang J. A. 2015. A discrete time Geom/G/1 retrial queue with balking customers and second optional service. OPSEARCH 53:344-357.
Woodward M.E. 1994. Communication and Computer Networks: Modelling with Discrete- Time Queues, IEEE Computer Society Press, Los Alamitos, CA.
Wu J.B., Wang J.X., and Liu Z.M. 2013. A discrete time Geom/G/1 retrial queue with preferred and impatient customers. Applied Mathematical Modeling. 37:2552-2561.