Abstract
The superposition of two independent Markovian arrival processes (MAPs) is also a Markovian arrival process of which the Markovian representation is given as the Kronecker sum of the transition rate matrices of the component processes. The moments of stationary intervals of the superposition can be obtained by differentiating the Laplace transform (LT) given in terms of the transition rate matrices. In this paper, we propose a streamlined procedure to determine the minimal LT of the merged process in terms of the minimal LT coefficients of the component processes. Combined with the closed-form transformation between moments and LT coefficients, our result enables us to determine the moments of the superposed process based on the moments of the component processes. The main contribution is that the whole procedure can be implemented without explicit Markovian representations. In order to transform the minimal LT coefficients of the component processes into the minimal LT representation of the merged process, we also introduce another minimal representation. A numerical example is provided to illustrate the procedure.
Similar content being viewed by others
References
Albin, S. L. (1982). On Poisson approximations for superposition arrival processes in queues. Management Science, 28(2), 126–137.
Bodrog, L., Heindl, A., Horváth, G., & Telek, M. (2008). A Markovian canonical form of second-order matrix-exponential processes. European Journal of Operational Research, 190(2), 459–477.
Bodrog, L., Horváth, G., & Telek, M. (2008). Moment characterization of matrix exponential and Markovian arrival processes. Annals of Operations Research, 160(1), 51–68.
Casale, G., Zhang, E., & Smirni, E. (2010). Trace data characterization and fitting for Markov modeling. Performance Evaluation, 67(2), 61–79.
Gantmacher, F. R. (1960). The theory of matrices. New York: Chelsea Publishing.
Horváth, A., Horváth, G., & Telek, M. (2010). A joint moments based analysis of networks of MAP/MAP/1 queues. Performance Evaluation, 67(9), 759–778.
Kang, S. H., Kim, Y. H., Sung, D. K., & Choi, B. D. (2002). An application of Markovian arrival process (MAP) to modeling superposed ATM cell streams. IEEE Transactions on Communications, 50(4), 633–642.
Kim, S. (2004). The heavy-traffic bottleneck phenomenon under splitting and superposition. European Journal of Operational Research, 157, 736–745.
Kim, S. (2011). Modeling cross correlation in three-moment four-parameter decomposition approximation of queueing networks. Operations Research, 59(2), 480–497.
Kim, S. (2017). Minimal LST representations of MAP(\(n\))s: Moment fittings and queueing approximations. Naval Research Logistics, 63(7), 549–561.
Kim, S., Muralidharan, R., & O’Cinneide, C. (2005). Taking account of correlations between streams in queueing network approximations. Queueing Systems, 49(5), 261–281.
Sriram, K., & Whitt, W. (1986). Characterizing superposition arrival processes in packet multiplexers for voice and data. IEEE Journal on Selected Areas in Communications, 4(6), 833–846.
Suresh, S., & Whitt, W. (1990). Arranging queues in series: A simulation experiment. Management Science, 36, 1080–1091.
Suresh, S., & Whitt, W. (1990). The heavy-traffic bottleneck phenomenon in open queueing networks. Operations Research Letters, 9, 355–362.
Telek, M., & Horváth, G. (2007). A minimal representation of Markov arrival processes and a moments matching method. Performance Evaluation, 64(9–12), 1153–1168.
Whitt, W. (1983). The queueing network analyzer. The Bell System Technical Journal, 62(9), 2779–2815.
Whitt, W. (1985). Queues with superposition arrival processes in heavy traffic. Stochastic Processes and their Applications, 21(1), 81–91.
Whitt, W., & You, W. (2021). A robust queueing network analyzer based on indices of dispersion. Naval Research Logistics, 69(1), 36–56.
Acknowledgements
This work was supported by the National Research Foundation of Korea Grant funded by the Korean Government (NRF-2017S1A5A2A01023654) and by the Ajou University research fund.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
1.1 Proof of Lemma 1
Proof
Let \(\tilde{f}(s) \equiv h(s)/g(s)\). Then, by Leibniz’s rule, the k-th order derivative of \(\tilde{f}(s) g(s)= h(s) \) is given as
by which we have
and the result follows by Eqs. (2.5) and (3.1).\(\square \)
1.2 Proof of Corollary 1
Proof
In order to determine \((\varvec{a}, \varvec{b})\), we need \(2n-1\) independent equations. For \(k = 1, 2,\ldots ,n-1\), Eq. (3.2) can be written as \(\textbf{R}_2 \varvec{a} =\varvec{b}\). For \(k = n, n+1,\ldots ,2n-1\), Eq. (3.2) can be written as
or in matrix form
where \(r_0 =1\). By multiplying every other equation by (\(-1\)) starting from the first one, we have \( \textbf{R}_1 \varvec{a} = \varvec{r}_{0..n-1}\). That is, for \(1 \le k \le 2n-1\), Eq. (3.2) can be written
\(\square \)
1.3 Proof of Lemma 2
Proof
Let \(w(s,t) = \sum _{i=1}^{n-1}\sum _{j=1}^{n-1} c_{i,j} s^i t^j \), \( h(s) =\sum _{i=1}^{n-1} b_{i} s^i+ a_0 \), \(g(s)=s^n +\sum _{i=0}^{n-1} a_i s^i\). Then, the joint LT in Eq. (2.12) can be written as
Let \(w_{k, l}(s,t) = \partial ^{k+l} w(s,t)/( \partial s^k \partial t^l)\) for which we have
By Leibniz’s rule with respect to s and t, it can be shown that
by which
and the result follows by Eqs. (2.6) and (3.1). \(\square \)
1.4 Proof of Lemma 3
Proof
Note that
Let \(\varvec{e}^\otimes = \varvec{e}_m \otimes \varvec{e}_n\). The subscript of \(\varvec{e}_m\) and \(\varvec{e}_n\) shall be suppressed below. Note that \(\hat{\textbf{D}}_0 \varvec{e}= -\hat{\textbf{D}}_1 \varvec{e} \) since \(\hat{\textbf{Q}} \varvec{e} = (\hat{\textbf{D}}_0+\hat{\textbf{D}}_1) \varvec{e} = \textbf{0}\). By Eq. (2.2) and the definition of \(\hat{y}_i\),
Likewise, \(\check{\varvec{p}}\check{\textbf{D}}_0^i \varvec{e} = - \check{y}_{i-1}\), \(\check{\varvec{\pi }}\check{\textbf{D}}_0^i \varvec{e} = \check{\lambda }_A \check{y}_{i-2}\), and \(\check{\varvec{\pi }}\check{\textbf{D}}_0^i \check{\textbf{D}}_1 \varvec{e} = - \check{\lambda }_A \check{y}_{i-1}\). By these identities and the definition of \(\hat{y}_k\) and \(\check{y}_k\),
Let \(\left( {\begin{array}{c}n\\ i\end{array}}\right) = 0 \) for \(i<0\) or \(i> n\). Then, by Pascal’s identity,
Likewise,
By Eqs. (5.1), (A.1), (A.2), (A.3), (A.4), and (A.5),
where the last equality is due to the same argument used in (A.4). \(\square \)
1.5 Proof of Lemma 4
Proof
By Eq. (A.1),
By Eq. (2.2) and the definition of \(\hat{z}_{ij}\),
Likewise, \(\check{\varvec{p}} \check{\textbf{D}}_0^i \check{\textbf{D}}_1 \check{\textbf{D}}_0^j\varvec{e} =-z_{i,j-1}\), \(\check{\varvec{\pi }}\check{\textbf{D}}_0^i \varvec{e} = \lambda _A y_{i-2}\), \(\check{\varvec{\pi }}\check{\textbf{D}}_0^i \check{\textbf{D}}_1 \check{\textbf{D}}_0^j\varvec{e} = \lambda _A z_{i-1,j-1}\), and \(\check{\varvec{\pi }}\check{\textbf{D}}_0^i \check{\textbf{D}}_1 \check{\textbf{D}}_0^j \check{\textbf{D}}_1 \varvec{e} = -\lambda _A z_{i-1,j}\). By these identities and the definition of \(\hat{y}_k\), \(\check{y}_k\), \(\hat{z}_{ij}\), and \(\check{z}_{ij}\), we have
Then, by the same argument as used in (A.4) with Pascal’s identity,
\(\square \)
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Kim, S. The superposition of Markovian arrival processes: moments and the minimal Laplace transform. Ann Oper Res 335, 237–259 (2024). https://doi.org/10.1007/s10479-024-05851-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10479-024-05851-7