Algorithms to determine the threshold reliability of flow networks

The threshold reliability R d of a flow network G can be defined in terms of the flow threshold d as the probability that network G admits a flow of at least d from the source node to the sink node. A fast exact algorithm to evaluate the threshold reliability of flow networks in terms of a sum of di...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IIE transactions Ročník 36; číslo 5; s. 469 - 479
Hlavní autoři: JANE, CHIN-CHIA, Laih, Yih-Wenn
Médium: Journal Article
Jazyk:angličtina
Vydáno: Norcross Taylor & Francis Group 01.05.2004
Taylor & Francis Ltd
Témata:
Algorithms
Approximation
Mathematical models
Operations research
Reliability
Studies
ISSN:0740-817X, 2472-5854, 1545-8830, 2472-5862
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:The threshold reliability R d of a flow network G can be defined in terms of the flow threshold d as the probability that network G admits a flow of at least d from the source node to the sink node. A fast exact algorithm to evaluate the threshold reliability of flow networks in terms of a sum of disjoint products is presented. In the proposed algorithm, a reliability product can be derived under a branching product in O(m) time using a d-path where m is the number of arcs in the network. Two algorithms are also presented to approximate the threshold reliability by modifying the proposed exact algorithm so that it can be terminated faster but at the cost of a lower accuracy. One is obtained by setting an acceptable difference value between the upper and lower bounds of the threshold reliability as the stopping criterion of the exact algorithm and the other is obtained by searching qualified α-products. Since the threshold reliability problem is NP-hard, approximation algorithms are indispensable for solving moderate and large flow networks. The proposed exact algorithm is verified using experimental data and the performances and properties of the two approximation algorithms are also explored and compared.
Bibliografie:SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 14
ISSN:0740-817X
2472-5854
1545-8830
2472-5862
DOI:10.1080/07408170490426170