The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application

Michał Witold Przewoźniczek, Krzysztof Walkowiak, Arunabha Sen, Marcin Komarnicki, Piotr Lechowicz

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Evolutionary methods are well-known tools used for solving hard computational problems. In this paper, we consider k-Shortest Steiner Trees (kSST) problem appearing in a diverse set of domains, e.g., multicast tree construction in communication networks in general, and optical networks in particular. The kSST is relatively new and has not been widely investigated in the literature. Thus, only a few algorithms have been proposed, each requiring significant resources amount and long execution times, partially following from the NP-hard nature of the problem. The kSST problem solution is a set of different trees, where each single tree can be easily represented using a genotype-encoding. However, encoding the tree set may be challenging, as each tree must be unique. Especially, in most applications the number of trees is large, resulting with the genotype containing high number of necessary genes. Thus, in this paper, we propose a transformation of the kSST problem into a dynamic problem, and when applied in the evolutionary method, a single individual encodes a single tree instead of a whole tree set. The results of extensive numerical experiments executed on two representative network topologies show that the proposed transformation improves the effectiveness of all considered state-of-the-art evolutionary methods.

Original languageEnglish (US)
Pages (from-to)1-19
Number of pages19
JournalInformation Sciences
Volume479
DOIs
StatePublished - Apr 1 2019

Fingerprint

Steiner Tree
Search Problems
Dynamic Problem
Fiber optic networks
Telecommunication networks
Genes
Topology
Binary
Steiner Tree Problem
Experiments
Genotype
Encoding
Evolutionary
Optical Networks
Multicast
Communication Networks
Network Topology
Execution Time
NP-complete problem
Numerical Experiment

Keywords

  • Dynamic problems
  • Dynamic subpopulation number control
  • Island models
  • K shortest Steiner trees
  • Linkage Learning
  • Parameter-less population pyramid

ASJC Scopus subject areas

  • Software
  • Control and Systems Engineering
  • Theoretical Computer Science
  • Computer Science Applications
  • Information Systems and Management
  • Artificial Intelligence

Cite this

The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application. / Przewoźniczek, Michał Witold; Walkowiak, Krzysztof; Sen, Arunabha; Komarnicki, Marcin; Lechowicz, Piotr.

In: Information Sciences, Vol. 479, 01.04.2019, p. 1-19.

Research output: Contribution to journalArticle

Przewoźniczek, Michał Witold ; Walkowiak, Krzysztof ; Sen, Arunabha ; Komarnicki, Marcin ; Lechowicz, Piotr. / The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application. In: Information Sciences. 2019 ; Vol. 479. pp. 1-19.
@article{713ee71ae21c4a38825539dadd55eb56,
title = "The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application",
abstract = "Evolutionary methods are well-known tools used for solving hard computational problems. In this paper, we consider k-Shortest Steiner Trees (kSST) problem appearing in a diverse set of domains, e.g., multicast tree construction in communication networks in general, and optical networks in particular. The kSST is relatively new and has not been widely investigated in the literature. Thus, only a few algorithms have been proposed, each requiring significant resources amount and long execution times, partially following from the NP-hard nature of the problem. The kSST problem solution is a set of different trees, where each single tree can be easily represented using a genotype-encoding. However, encoding the tree set may be challenging, as each tree must be unique. Especially, in most applications the number of trees is large, resulting with the genotype containing high number of necessary genes. Thus, in this paper, we propose a transformation of the kSST problem into a dynamic problem, and when applied in the evolutionary method, a single individual encodes a single tree instead of a whole tree set. The results of extensive numerical experiments executed on two representative network topologies show that the proposed transformation improves the effectiveness of all considered state-of-the-art evolutionary methods.",
keywords = "Dynamic problems, Dynamic subpopulation number control, Island models, K shortest Steiner trees, Linkage Learning, Parameter-less population pyramid",
author = "Przewoźniczek, {Michał Witold} and Krzysztof Walkowiak and Arunabha Sen and Marcin Komarnicki and Piotr Lechowicz",
year = "2019",
month = "4",
day = "1",
doi = "10.1016/j.ins.2018.11.015",
language = "English (US)",
volume = "479",
pages = "1--19",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - The transformation of the k-Shortest Steiner trees search problem into binary dynamic problem for effective evolutionary methods application

AU - Przewoźniczek, Michał Witold

AU - Walkowiak, Krzysztof

AU - Sen, Arunabha

AU - Komarnicki, Marcin

AU - Lechowicz, Piotr

PY - 2019/4/1

Y1 - 2019/4/1

N2 - Evolutionary methods are well-known tools used for solving hard computational problems. In this paper, we consider k-Shortest Steiner Trees (kSST) problem appearing in a diverse set of domains, e.g., multicast tree construction in communication networks in general, and optical networks in particular. The kSST is relatively new and has not been widely investigated in the literature. Thus, only a few algorithms have been proposed, each requiring significant resources amount and long execution times, partially following from the NP-hard nature of the problem. The kSST problem solution is a set of different trees, where each single tree can be easily represented using a genotype-encoding. However, encoding the tree set may be challenging, as each tree must be unique. Especially, in most applications the number of trees is large, resulting with the genotype containing high number of necessary genes. Thus, in this paper, we propose a transformation of the kSST problem into a dynamic problem, and when applied in the evolutionary method, a single individual encodes a single tree instead of a whole tree set. The results of extensive numerical experiments executed on two representative network topologies show that the proposed transformation improves the effectiveness of all considered state-of-the-art evolutionary methods.

AB - Evolutionary methods are well-known tools used for solving hard computational problems. In this paper, we consider k-Shortest Steiner Trees (kSST) problem appearing in a diverse set of domains, e.g., multicast tree construction in communication networks in general, and optical networks in particular. The kSST is relatively new and has not been widely investigated in the literature. Thus, only a few algorithms have been proposed, each requiring significant resources amount and long execution times, partially following from the NP-hard nature of the problem. The kSST problem solution is a set of different trees, where each single tree can be easily represented using a genotype-encoding. However, encoding the tree set may be challenging, as each tree must be unique. Especially, in most applications the number of trees is large, resulting with the genotype containing high number of necessary genes. Thus, in this paper, we propose a transformation of the kSST problem into a dynamic problem, and when applied in the evolutionary method, a single individual encodes a single tree instead of a whole tree set. The results of extensive numerical experiments executed on two representative network topologies show that the proposed transformation improves the effectiveness of all considered state-of-the-art evolutionary methods.

KW - Dynamic problems

KW - Dynamic subpopulation number control

KW - Island models

KW - K shortest Steiner trees

KW - Linkage Learning

KW - Parameter-less population pyramid

UR - http://www.scopus.com/inward/record.url?scp=85057168913&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85057168913&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2018.11.015

DO - 10.1016/j.ins.2018.11.015

M3 - Article

VL - 479

SP - 1

EP - 19

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -