A novel approach for fitting probability distributions to real trace data with the em algorithm

Axel Thümmler, Peter Buchholz, M. Telek

Research output: Chapter in Book/Report/Conference proceedingConference contribution

27 Citations (Scopus)

Abstract

The representation of general distributions or measured data by phase-type distributions is an important and non-trivial task in analytical modeling. Although a large number of different methods for fitting parameters of phase-type distributions to data traces exist, many approaches lack efficiency and numerical stability. In this paper, a novel approach is presented that fits a restricted class of phase-type distributions, namely mixtures of Erlang distributions, to trace data. For the parameter fitting an algorithm of the expectation maximization type is developed. The paper shows that these choices result in a very efficient and numerically stable approach which yields phase-type approximations for a wide range of data traces that are as good or better than approximations computed with other less efficient and less stable fitting methods. To illustrate the effectiveness of the proposed fitting algorithm, we present comparative results for our approach and two other methods using six benchmark traces and two real traffic traces.

Original languageEnglish
Title of host publicationProceedings of the International Conference on Dependable Systems and Networks
Pages712-721
Number of pages10
Publication statusPublished - 2005
Event2005 International Conference on Dependable Systems and Networks - Yokohama, Japan
Duration: Jun 28 2005Jul 1 2005

Other

Other2005 International Conference on Dependable Systems and Networks
CountryJapan
CityYokohama
Period6/28/057/1/05

Fingerprint

Probability distributions
Convergence of numerical methods

ASJC Scopus subject areas

  • Computer Science (miscellaneous)
  • Computer Networks and Communications

Cite this

Thümmler, A., Buchholz, P., & Telek, M. (2005). A novel approach for fitting probability distributions to real trace data with the em algorithm. In Proceedings of the International Conference on Dependable Systems and Networks (pp. 712-721)

A novel approach for fitting probability distributions to real trace data with the em algorithm. / Thümmler, Axel; Buchholz, Peter; Telek, M.

Proceedings of the International Conference on Dependable Systems and Networks. 2005. p. 712-721.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Thümmler, A, Buchholz, P & Telek, M 2005, A novel approach for fitting probability distributions to real trace data with the em algorithm. in Proceedings of the International Conference on Dependable Systems and Networks. pp. 712-721, 2005 International Conference on Dependable Systems and Networks, Yokohama, Japan, 6/28/05.
Thümmler A, Buchholz P, Telek M. A novel approach for fitting probability distributions to real trace data with the em algorithm. In Proceedings of the International Conference on Dependable Systems and Networks. 2005. p. 712-721
Thümmler, Axel ; Buchholz, Peter ; Telek, M. / A novel approach for fitting probability distributions to real trace data with the em algorithm. Proceedings of the International Conference on Dependable Systems and Networks. 2005. pp. 712-721
@inproceedings{8f499bf927b54bfba51908fd926e6834,
title = "A novel approach for fitting probability distributions to real trace data with the em algorithm",
abstract = "The representation of general distributions or measured data by phase-type distributions is an important and non-trivial task in analytical modeling. Although a large number of different methods for fitting parameters of phase-type distributions to data traces exist, many approaches lack efficiency and numerical stability. In this paper, a novel approach is presented that fits a restricted class of phase-type distributions, namely mixtures of Erlang distributions, to trace data. For the parameter fitting an algorithm of the expectation maximization type is developed. The paper shows that these choices result in a very efficient and numerically stable approach which yields phase-type approximations for a wide range of data traces that are as good or better than approximations computed with other less efficient and less stable fitting methods. To illustrate the effectiveness of the proposed fitting algorithm, we present comparative results for our approach and two other methods using six benchmark traces and two real traffic traces.",
author = "Axel Th{\"u}mmler and Peter Buchholz and M. Telek",
year = "2005",
language = "English",
pages = "712--721",
booktitle = "Proceedings of the International Conference on Dependable Systems and Networks",

}

TY - GEN

T1 - A novel approach for fitting probability distributions to real trace data with the em algorithm

AU - Thümmler, Axel

AU - Buchholz, Peter

AU - Telek, M.

PY - 2005

Y1 - 2005

N2 - The representation of general distributions or measured data by phase-type distributions is an important and non-trivial task in analytical modeling. Although a large number of different methods for fitting parameters of phase-type distributions to data traces exist, many approaches lack efficiency and numerical stability. In this paper, a novel approach is presented that fits a restricted class of phase-type distributions, namely mixtures of Erlang distributions, to trace data. For the parameter fitting an algorithm of the expectation maximization type is developed. The paper shows that these choices result in a very efficient and numerically stable approach which yields phase-type approximations for a wide range of data traces that are as good or better than approximations computed with other less efficient and less stable fitting methods. To illustrate the effectiveness of the proposed fitting algorithm, we present comparative results for our approach and two other methods using six benchmark traces and two real traffic traces.

AB - The representation of general distributions or measured data by phase-type distributions is an important and non-trivial task in analytical modeling. Although a large number of different methods for fitting parameters of phase-type distributions to data traces exist, many approaches lack efficiency and numerical stability. In this paper, a novel approach is presented that fits a restricted class of phase-type distributions, namely mixtures of Erlang distributions, to trace data. For the parameter fitting an algorithm of the expectation maximization type is developed. The paper shows that these choices result in a very efficient and numerically stable approach which yields phase-type approximations for a wide range of data traces that are as good or better than approximations computed with other less efficient and less stable fitting methods. To illustrate the effectiveness of the proposed fitting algorithm, we present comparative results for our approach and two other methods using six benchmark traces and two real traffic traces.

UR - http://www.scopus.com/inward/record.url?scp=27544489505&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=27544489505&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:27544489505

SP - 712

EP - 721

BT - Proceedings of the International Conference on Dependable Systems and Networks

ER -