期刊
IEEE ACCESS
卷 10, 期 -, 页码 127589-127600出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2022.3226687
关键词
Disjoint path; path cover; path partition; torus; toroidal grid; interconnection network
资金
- National Research Foundation of Korea (NRF) - Korean Government (MSIT) [2021R1F1A1048180]
- Catholic University of Korea
- National Research Foundation of Korea [2021R1F1A1048180] Funding Source: Korea Institute of Science & Technology Information (KISTI), National Science & Technology Information Service (NTIS)
This paper discusses the problem of finding disjoint paths in the underlying graph of an interconnection network in parallel processing. It reveals that constructing a nonbipartite torus-like graph with good disjoint-path-cover properties from lower dimensional torus-like graphs can retain the good property. Consequently, nonbipartite tori with at most f vertex and/or edge faults can have unpaired many-to-many disjoint path covers joining arbitrary disjoint sets.
One of the key problems in parallel processing is finding disjoint paths in the underlying graph of an interconnection network. The disjoint path cover of a graph is a set of pairwise vertex-disjoint paths that altogether cover every vertex of the graph. Given disjoint source and sink sets, S = {s(1), ... , s(k)} and T = {t(1), ... , t(k)}, in graph G, an unpaired many-to-many k-disjoint path cover joining S and T is a disjoint path cover {P-1, ... , P-k}, in which each path P-i runs from source s(i) to some sink t(j). In this paper, we reveal that a nonbipartite torus-like graph, if built from lower dimensional torus-like graphs that have good disjoint-path-cover properties of the unpaired type, retains such a good property. As a result, an m-dimensional nonbipartite torus, m >= 2, with at most f vertex and/or edge faults has an unpaired many-to-many k-disjoint path cover joining arbitrary disjoint sets S and T of size k each, subject to k >= 1 and f + k <= 2m - 2. The bound of 2m - 2 on f + k is nearly optimal.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据