Closed form solutions for mapping general distributions to quasi-minimal PH distributions

Approximating general distributions by phase-type (PH) distributions is a popular technique in stochastic analysis, since the Markovian property of PH distributions often allows analytical tractability. This paper proposes an algorithm for mapping a general distribution, G, to a PH distribution, which matches the first three moments of G. Efficiency of our algorithm hinges on narrowing the search space to a particular subset of the PH distributions, which we refer to as Erlang-Coxian (EC) distributions. The class of EC distributions has a small number of parameters, and we provide closed form solutions for these. Our solution applies to any distribution whose first three moments can be matched by a PH distribution. Also, our resulting EC distribution requires a nearly minimal number of phases, within one of the minimal number of phases required by any acyclic PH distribution.

[1]  L. Schmickler.  Meda: mixed erlang distributions as phase-type representations of empirical distribution functions. , 1992 .

[2]  Ward Whitt,et al.  Approximating a point process by a renewal process , 1981 .

[3]  R. F. Brown,et al.  PERFORMANCE EVALUATION , 2019, ISO 22301:2019 and business continuity management – Understand how to plan, implement and enhance a business continuity management system (BCMS).

[4]  Moshe Sidi,et al.  Modeling and analysis of power-tail distributions via classical teletraffic methods , 2000, Queueing Syst. Theory Appl..

[5]  Mor Harchol-Balter,et al.  Necessary and Sufficient Conditions for Representing General Distributions by Coxians , 2003, Computer Performance Evaluation / TOOLS.

[6]  Marcel F. Neuts,et al.  Matrix-geometric solutions in stochastic models - an algorithmic approach , 1982 .

[7]  M. A. Johnson,et al.  Selecting Parameters of Phase Distributions: Combining Nonlinear Programming, Heuristics, and Erlang Distributions , 1993, INFORMS J. Comput..

[8]  K. Mani Chandy,et al.  Approximate Analysis of Central Server Models , 1975, IBM J. Res. Dev..

[9]  M. A. Johnson,et al.  An investigation of phase-distribution moment-matching algorithms for use in queueing models , 1991, Queueing Syst. Theory Appl..

[10]  M. A. Johnson,et al.  A graphical investigation of error bounds for moment-based queueing approximations , 1991, Queueing Syst. Theory Appl..

[11]  MarieRaymond.  Calculating equilibrium probabilities for (n)/Ck/1/N queues , 1980 .

[12]  Mor Harchol-Balter,et al.  Analysis of multi-server systems via dimensionality reduction of markov chains , 2005 .

[13]  T. Altiok.  On the Phase-Type Approximations of General Distributions , 1985 .

[14]  Mark S. Squillante,et al.  Analysis of task assignment with cycle stealing under central queue , 2003, 23rd International Conference on Distributed Computing Systems, 2003. Proceedings..

[15]  Anand Sivasubramaniam,et al.  An Integrated Approach to Parallel Scheduling Using Gang-Scheduling, Backfilling, and Migration , 2001, JSSPP.

[16]  Colm Art O'Cinneide,et al.  Phase-Type Distributions and Majorization , 1991 .

[17]  W. Zijm,et al.  Computational procedures for stochastic multi-echelon production systems , 1991 .

[18]  A. David,et al.  The least variable phase type distribution is Erlang , 1987 .

[19]  M. A. Johnson,et al.  Matching moments to phase distributions: nonlinear programming approaches , 1990 .

[20]  Mor Harchol-Balter,et al.  A Closed-Form Solution for Mapping General Distributions to Minimal PH Distributions , 2003, Computer Performance Evaluation / TOOLS.

[21]  Miklós Telek,et al.  PhFit: a general phase-type fitting tool , 2002, Proceedings International Conference on Dependable Systems and Networks.

[22]  A. Horváth,et al.  Approximating heavy tailed behaviour with Phase type distributions , 2000 .

[23]  H. Stadtler,et al.  A comparison of two optimization procedures for 1- and 1 1/2-dimensional cutting stock problems , 1988 .

[24]  Attahiru Sule Alfa,et al.  Advances in matrix-analytic methods for stochastic models , 1998 .

[25]  W. J. Studden,et al.  Tchebycheff Systems: With Applications in Analysis and Statistics. , 1967 .

[26]  Raymond A. Marie,et al.  Calculating equilibrium probabilities for &lgr;(n)/Ck/1/N queues , 1980 .

[27]  A. Cumani.  On the canonical representation of homogeneous markov processes modelling failure - time distributions , 1982 .

[28]  Michael A. Johnson,et al.  Matching moments to phase distri-butions: mixtures of Erlang distribution of common order , 1989 .

[29]  Anja Feldmann,et al.  Fitting mixtures of exponentials to long-tail distributions to analyze network performance models , 1997, Proceedings of INFOCOM '97.

[30]  M. A. Johnson,et al.  Matching moments to phase distributions: density function shapes , 1990 .

[31]  Alma Riska,et al.  An EM-based technique for approximating long-tailed data sets with PH distributions , 2004, Perform. Evaluation.

[32]  W. Whitt.  On approximations for queues, III: Mixtures of exponential distributions , 1984, AT&T Bell Laboratories Technical Journal.

[33]  Bruno Sericola,et al.  Transient Analysis of Stochastic Fluid Models , 1998, Perform. Evaluation.

[34]  Miklós Telek,et al.  PhFit: A General Phase-Type Fitting Tool , 2002, Computer Performance Evaluation / TOOLS.

[35]  J. Moreira,et al.  An Evaluation of Parallel Job Scheduling for ASCI Blue-Pacific , 1999, ACM/IEEE SC 1999 Conference (SC'99).

[36]  Alan Scheller-Wolf,et al.  Analysis of cycle stealing with switching cost , 2003, SIGMETRICS '03.

[37]  Ramin Sadre,et al.  Fitting World Wide Web request traces with the EM-algorithim , 2001, SPIE ITCom.

[38]  G. J. Houtum,et al.  Incomplete convolutions in production and inventory models , 1997 .