Reliability assessment of multiprocessor system based on (n, k)-star network

Shuming Zhou, Xiaowang Li, Jinqiang Li, Dajin Wang

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)

Abstract

As the size and complexity of a multiprocessor system increases, reliability evaluation becomes an important issue. The performability of a multiprocessor system heavily depends on the application program and the underlying architecture. In multitasking multiprocessor system, the problem of dynamically assigning a given dimensional subsystem to a special task is considered as a reallocation in the presence of node and/or link failures. This paper takes the generalization of star graph, (n, k)-star graph, as an empirical object. In order to measure the reliability of (n, k)star graph, the analytical model introduces mean time to failure (MTTF) to show the time that the appearance of a certain number of faulty S n −1 ,k− 1 costs. The higher the MTTF, the better the robustness. So, the way to evaluate the robustness of an (n, k)-star is to count how much the MTTF is. In fact, an (n, k)-star can be partitioned along any dimension (except the first one) with corresponding identification code. So, we will explore the reliability of (n, k)-star graph when it is partitioned along any dimension (except the first one) under node and/or link fault model. Comparisons among the simulation results under two partitioning models reveal that the MTTF is higher under liberal partition model, which better reflect the steady state of an interconnection network that can persist when the network is destroyed.

Original languageEnglish
Pages (from-to)1025-1035
Number of pages11
JournalIEEE Transactions on Reliability
Volume66
Issue number4
DOIs
StatePublished - 1 Dec 2017

Fingerprint

Stars
Multitasking
Application programs
Analytical models
Costs

Keywords

  • (n,k)-star
  • Mean time to failure (MTTF)
  • Node (link) failure
  • Reliability
  • Robustness

Cite this

Zhou, Shuming ; Li, Xiaowang ; Li, Jinqiang ; Wang, Dajin. / Reliability assessment of multiprocessor system based on (n, k)-star network. In: IEEE Transactions on Reliability. 2017 ; Vol. 66, No. 4. pp. 1025-1035.
@article{ce40f3b66cb2477582a7ea15cf697429,
title = "Reliability assessment of multiprocessor system based on (n, k)-star network",
abstract = "As the size and complexity of a multiprocessor system increases, reliability evaluation becomes an important issue. The performability of a multiprocessor system heavily depends on the application program and the underlying architecture. In multitasking multiprocessor system, the problem of dynamically assigning a given dimensional subsystem to a special task is considered as a reallocation in the presence of node and/or link failures. This paper takes the generalization of star graph, (n, k)-star graph, as an empirical object. In order to measure the reliability of (n, k)star graph, the analytical model introduces mean time to failure (MTTF) to show the time that the appearance of a certain number of faulty S n −1 ,k− 1 costs. The higher the MTTF, the better the robustness. So, the way to evaluate the robustness of an (n, k)-star is to count how much the MTTF is. In fact, an (n, k)-star can be partitioned along any dimension (except the first one) with corresponding identification code. So, we will explore the reliability of (n, k)-star graph when it is partitioned along any dimension (except the first one) under node and/or link fault model. Comparisons among the simulation results under two partitioning models reveal that the MTTF is higher under liberal partition model, which better reflect the steady state of an interconnection network that can persist when the network is destroyed.",
keywords = "(n,k)-star, Mean time to failure (MTTF), Node (link) failure, Reliability, Robustness",
author = "Shuming Zhou and Xiaowang Li and Jinqiang Li and Dajin Wang",
year = "2017",
month = "12",
day = "1",
doi = "10.1109/TR.2017.2740339",
language = "English",
volume = "66",
pages = "1025--1035",
journal = "IEEE Transactions on Reliability",
issn = "0018-9529",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "4",

}

Reliability assessment of multiprocessor system based on (n, k)-star network. / Zhou, Shuming; Li, Xiaowang; Li, Jinqiang; Wang, Dajin.

In: IEEE Transactions on Reliability, Vol. 66, No. 4, 01.12.2017, p. 1025-1035.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Reliability assessment of multiprocessor system based on (n, k)-star network

AU - Zhou, Shuming

AU - Li, Xiaowang

AU - Li, Jinqiang

AU - Wang, Dajin

PY - 2017/12/1

Y1 - 2017/12/1

N2 - As the size and complexity of a multiprocessor system increases, reliability evaluation becomes an important issue. The performability of a multiprocessor system heavily depends on the application program and the underlying architecture. In multitasking multiprocessor system, the problem of dynamically assigning a given dimensional subsystem to a special task is considered as a reallocation in the presence of node and/or link failures. This paper takes the generalization of star graph, (n, k)-star graph, as an empirical object. In order to measure the reliability of (n, k)star graph, the analytical model introduces mean time to failure (MTTF) to show the time that the appearance of a certain number of faulty S n −1 ,k− 1 costs. The higher the MTTF, the better the robustness. So, the way to evaluate the robustness of an (n, k)-star is to count how much the MTTF is. In fact, an (n, k)-star can be partitioned along any dimension (except the first one) with corresponding identification code. So, we will explore the reliability of (n, k)-star graph when it is partitioned along any dimension (except the first one) under node and/or link fault model. Comparisons among the simulation results under two partitioning models reveal that the MTTF is higher under liberal partition model, which better reflect the steady state of an interconnection network that can persist when the network is destroyed.

AB - As the size and complexity of a multiprocessor system increases, reliability evaluation becomes an important issue. The performability of a multiprocessor system heavily depends on the application program and the underlying architecture. In multitasking multiprocessor system, the problem of dynamically assigning a given dimensional subsystem to a special task is considered as a reallocation in the presence of node and/or link failures. This paper takes the generalization of star graph, (n, k)-star graph, as an empirical object. In order to measure the reliability of (n, k)star graph, the analytical model introduces mean time to failure (MTTF) to show the time that the appearance of a certain number of faulty S n −1 ,k− 1 costs. The higher the MTTF, the better the robustness. So, the way to evaluate the robustness of an (n, k)-star is to count how much the MTTF is. In fact, an (n, k)-star can be partitioned along any dimension (except the first one) with corresponding identification code. So, we will explore the reliability of (n, k)-star graph when it is partitioned along any dimension (except the first one) under node and/or link fault model. Comparisons among the simulation results under two partitioning models reveal that the MTTF is higher under liberal partition model, which better reflect the steady state of an interconnection network that can persist when the network is destroyed.

KW - (n,k)-star

KW - Mean time to failure (MTTF)

KW - Node (link) failure

KW - Reliability

KW - Robustness

UR - http://www.scopus.com/inward/record.url?scp=85052756869&partnerID=8YFLogxK

U2 - 10.1109/TR.2017.2740339

DO - 10.1109/TR.2017.2740339

M3 - Article

VL - 66

SP - 1025

EP - 1035

JO - IEEE Transactions on Reliability

JF - IEEE Transactions on Reliability

SN - 0018-9529

IS - 4

ER -