Abstract
Motivated by experiments on customers’ behavior in service systems, we consider a queueing model with eventdependent arrival rates. Customers’ arrival rates depend on the last event, which may either be a service departure or an arrival. We derive explicitly the performance measures and analyze the impact of the eventdependency. In particular, we show that this queueing model, in which a service completion generates a higher arrival rate than an arrival, performs better than a system in which customers are insensitive to the last event. Moreover, contrary to the M/G/1 queue, we show that the coefficient of variation of the service does not necessarily deteriorate the system performance. Next, we show that this queueing model may be the result of customers’ strategic behavior when only the last event is known. Finally, we investigate the historical admission control problem. We show that, under certain conditions, a deterministic policy with two thresholds may be optimal. This new policy is easy to implement and provides an improvement compared to the classical onethreshold policy.
This is a preview of subscription content, access via your institution.
Notes
The performance measures in the exponential case could also be obtained using a Markov chain analysis.
Another way to compute the performance measures in the Erlang case is to use a matrixgeometric approach.
We refer the reader to the book [12] for an overview on equilibrium behavior in queueing systems.
In this case, \(E(W^+)=\overline{x}\frac{1+cv^2}{2}\left( \frac{1}{G^*(\lambda )}+\frac{\lambda \overline{x}}{1\lambda \overline{x}}\right) +\frac{1\lambda \overline{x}}{\lambda G^*(\lambda )}\frac{1}{\lambda }\), and \(E(W^)=\overline{x}\left( \frac{1+cv^2}{2}\frac{(\lambda \overline{x})^2(1G^*(\lambda ))}{(1\lambda \overline{x})(1\lambda \overline{x}G^*(\lambda ))}+\frac{G^*(\lambda )+\lambda \overline{x}1}{1\lambda \overline{x}G^*(\lambda )} \right) \).
The limit of \(E(W^+)\) corresponds to the expected remaining service time in an M/G/1 queue.
[3] shows an example where deterministic policies are not optimal for the admission control in an M/G/1 queue.
This particular Coxian distribution is a hypoexponential distribution.
Yet, we do not claim that the twothreshold policy is optimal in this case.
Abbreviations
 \(\lambda ^+\), \(\lambda ^\) :

Arrival rates after an arrival and after a service
 S :

Random variable which represents the service time duration
 \(\overline{x}\) :

Expected service time
 cv :

Coefficient of variation of the service time distribution; it is the ratio of the standard deviation divided by the expected value
 g(.):

Probability density function of the service time
 \(G^*(.)\) :

Laplace–Stieltjes transform (LST) of the service time; \(G^*(s)=\int _{0}^{\infty } g(t) e^{s t} \, {\mathrm {d}}t\)
 \(k^+\), \(k^\) :

Thresholds on the system size or on the remaining number of service phases to accept or reject customers at arrival after an arrival or a service
 \(\widetilde{p}^+\), \(\widetilde{p}^\) :

Probabilities that an arriving customer accepts joining after an arrival or a service
 \(p_t(n,r,+)\), \(p_t(n,r,)\) :

Probability density of having n customers in the system, \(n \ge 1\), and a remaining service time of r, \(r \ge 0\), at time t after an arrival or a service
 \(p(n,r,+)\), \(p(n,r,+)\) :

\(p(n,r,+)=\lim \limits _{t \rightarrow \infty } p_t(n,r,+)\), \(p(n,r,)=\lim \limits _{t \rightarrow \infty } p_t(n,r,)\) for \(n\ge 1\)
 \(\pi _{n,+}\), \(\pi _{n,}\) :

Stationary probability of having n customers in the system at arbitrary instants after an arrival or a service (\(\pi _{n,+}=\int _{0}^{\infty } p(n,r,+) \, {\mathrm {d}}r\), \(\pi _{n,}=\int _{0}^{\infty } p(n,r,) \, {\mathrm {d}}r\) for \(n\ge 1\))
 \(\pi _{n}\) :

Stationary probability of having n customers in the system at arbitrary instants, \(\pi _{n}=\pi _{n,+}+\pi _{n,}\) for \(n\ge 0\)
 \(p_{n,+}\), \(p_{n,}\) :

Stationary probability of having n customers in the system at departure instants after an arrival or a service for \(n\ge 0\)
 \(p_{n}\) :

Stationary probability of having n customers in the system at departure instants, \(p_{n}=p_{n,+}+p_{n,}\) for \(n\ge 0\)
 \(\overline{r_n^+}\), \(\overline{r_n^}\) :

Expected remaining service time seen by a customer who arrives after an arrival or a service with n customers present in the system, \(n\ge 1\).
 \(\alpha _{n}\), \(\beta _n\) :

Probability that n customers arrive during a service if the service is initiated by a service completion or by an arrival, \(n\ge 0\)
 \(E(Q_d)\), E(Q):

Expected number of customers in the system at departure and arbitrary instants
 \(\overline{E(Q)}\) :

Service level objective on E(Q)
 E(W):

Expected waiting time at arbitrary instants
 \(T_s\) :

Expected throughput of served customers
 \(E(W^+)\), \(E(W^)\) :

Expected waiting time of a customer who arrives after an arrival or a service
 E(R):

Expected remaining service time seen by an arriving customer at a nonempty system
 \(E(R^+)\), \(E(R^)\) :

Expected remaining service time seen by a customer who arrives after an arrival or a service
 \(V_n^+(x)\), \(V_n^(x)\) :

Value function depending on the state of the system
 \(\mu _j\) :

Exponential rate of remaining service phase j in the Coxian distribution (\(1\le j \le N\))
 \(r_j\) :

Probability of entering remaining service phase \(j1\) after leaving remaining service phase j (\(1\le j \le N\)) in the Coxian distribution
References
Aksin, Z.: The effect of random waits on customer queue joining and reneging behavior: a laboratory experiment. In: Proceedings of the EURO XXVII Annual Conference, Glasgow (2015)
Altman, E.: Constrained Markov Decision Processes, vol. 7. CRC Press, Boca Raton (1999)
Altman, E., Hassin, R.: Nonthreshold equilibrium for customers joining an M/G/1 queue. In: Proceedings of 10th International Symposium on Dynamic Game and Applications. Citeseer (2002)
Asmussen, S., Nerman, O., Olsson, M.: Fitting phasetype distributions via the EM algorithm. Scand. J. Stat. 1, 419–441 (1996)
Bekker, R., Borst, S.C., Boxma, O.J., Kella, O.: Queues with workloaddependent arrival and service rates. Queueing Syst. 46(3–4), 537–556 (2004)
Bekker, R., Koole, G.M., Nielsen, B.F., Nielsen, T.B.: Queues with waiting time dependent service. Queueing Syst. 68(1), 61–78 (2011)
Bellman, R.E.: Dynamic Programming. Princeton University Press, Princeton (1957)
Boxma, O.J.: Joint distribution of sojourn time and queue length in the M/G/1 queue with (in) finite capacity. Eur. J. Oper. Res. 16(2), 246–256 (1984)
Boxma, O.J., Kaspi, H., Kella, O., Perry, D.: On/off storage systems with statedependent input, output, and switching rates. Probab. Eng. Inf. Sci. 19(01), 1–14 (2005)
Boxma, O.J., Vlasiou, M.: On queues with service and interarrival times depending on waiting times. Queueing Syst. 56(3–4), 121–132 (2007)
Harrison, J.M., Resnick, S.I.: The stationary distribution and first exit probabilities of a storage process with general release rule. Math. Oper. Res. 1(4), 347–358 (1976)
Hassin, R., Haviv, M.: To Queue or Not to Queue: Equilibrium Behavior in Queueing Systems, vol. 59. Springer, New York (2003)
Horváth, A., Telek, M.: Phfit: a general phasetype fitting tool. Comput. Perform. Eval. Modell. Tech. Tools 1, 1–14 (2002)
Howard, R.A.: Dynamic programming and Markov processes. Wiley, New York (1960)
Kerner, Y.: The conditional distribution of the residual service time in the Mn/G/1 queue. Stoch. Models 24(3), 364–375 (2008)
Kleinrock, L.: Queueing Systems, Theory, vol. I. Wiley, New York (1975)
Koole, G.: A simple proof of the optimality of a threshold policy in a twoserver queueing system. Syst. Control Lett. 26(5), 301–303 (1995)
Koole, G.: Monotonicity in Markov Reward and Decision Chains: Theory and Applications, vol. 1. Now Publishers Inc, New York (2007)
Larsen, R.L., Agrawala, A.K.: Control of a heterogeneous twoserver exponential queueing system. IEEE Trans. Softw. Eng. 4, 522–526 (1983)
Legros, B., Jouini, O.: Routing in a queueing system with two heterogeneous servers in speed and in quality of resolution. Stoch. Models (2017). https://doi.org/10.1080/15326349.2017.1303615
Legros, B., Sezer, A.D.: Stationary analysis of a single queue with remaining service time dependent arrivals. Queueing Syst. (2017). https://doi.org/10.1007/s111340179552z
Lin, W., Kumar, P.R.: Optimal control of a queueing system with two heterogeneous servers. IEEE Trans. Autom. Control 29(8), 696–703 (1984)
Luh, H.P., Viniotis, I.: Threshold control policies for heterogeneous server systems. Math. Methods Oper. Res. 55(1), 121–142 (2002)
Neuts, M.F.: MatrixGeometric Solutions in Stochastic Models: An Algorithmic Approach. Johns Hopkins University Press, Mineola (1981)
Puterman, M.L.: Markov Decision Processes. Wiley, New York (1994)
Rykov, V.V.: Monotone control of queueing systems with heterogeneous servers. Queueing Syst. 37(4), 391–403 (2001)
Schassberger, R.: Warteschlangen. Verlag, Vienna (1973)
Sigman, K., Yechiali, U.: Stationary remaining service time conditional on queue length. Oper. Res Lett. 35(5), 581–583 (2007)
Van Der Heijden, M.C.: On the threemoment approximation of a general distribution by a Coxian distribution. Probab. Eng. Inf. Sci. 2(2), 257–261 (1988)
Walrand, J.: A note on optimal control of a queuing system with two heterogeneous servers. Syst. Control Lett. 4(3), 131–134 (1984)
Whitt, W.: Queues with service times and interarrival times depending linearly and randomly upon waiting times. Queueing Syst. 6(1), 335–351 (1990)
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
About this article
Cite this article
Legros, B. M/G/1 queue with eventdependent arrival rates. Queueing Syst 89, 269–301 (2018). https://doi.org/10.1007/s1113401795577
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s1113401795577
Keywords
 Queueing systems
 Performance evaluation
 M/G/1
 Threshold policy
 Strategic behavior
Mathematics Subject Classification
 90B22
 60K25
 68M20