Conditional fault diameter of crossed cubes

https://doi.org/10.1016/j.jpdc.2008.08.001Get rights and content

Abstract

The conditional connectivity and the conditional fault diameter of a crossed cube are studied in this work. The conditional connectivity is the connectivity of an interconnection network with conditional faults, where each node has at least one fault-free neighbor. Based on this requirement, the conditional connectivity of a crossed cube is shown to be 2n2. Extending this result, the conditional fault diameter of a crossed cube is also shown to be D(CQn)+3 as a set of 2n3 node failures. This indicates that the conditional fault diameter of a crossed cube is increased by three compared to the fault-free diameter of a crossed cube. The conditional fault diameter of a crossed cube is approximately half that of the hypercube. In this respect, the crossed cube is superior to the hypercube.

Introduction

Several studies have proposed and examined the feasibility topologies of multiprocessor interconnection networks. Such a topology is usually modeled as an undirected graph in which the set of nodes represents the processors and the set of edges represents the communication links between the processors. The fault tolerance of an interconnection network can be measured from the (node) connectivity of the underlying graph. In a network with connectivity of n, the fault-free node is guaranteed to communicate with any other fault-free node even if (n1) nodes are faulty. Furthermore, the fault diameter is a generalized measure to determine this fault tolerance aspect of the network. The fault diameter can estimate the effect on the diameter when faults occur. A small fault diameter is desirable to obtain a shorter delay in communication when node faults occur.

However, the connectivity is not an appropriate measure of fault tolerance for a network in which some subsets of the node do not fail at the same time. For example, the connectivity of a hypercube is n. The hypercube can be disconnected only by removing all of the neighbors of one node, which is very improbable. Motivated by this observation, Esfahanian [6] introduced the concept that all of the neighbors of each node cannot be faulty at the same time. Under this restriction, the hypercube can tolerate up to 2n3 faulty nodes without being disrupted. Therefore, the connectivity with conditional faults provides accurate measures of fault tolerance for a network. Moreover, a fault diameter with conditional faults represents a more realistic delay time in communication when node faults occur.

The concept of conditional faults is more general than that of the original one. The conditional connectivity measures for large multiprocessor systems have been investigated in [14]. Latifi [13] determined that the conditional fault diameter of the hypercube is n+2; the conditional fault diameter of star graph networks was considered in [17]. The conditional connectivity of k-ary n-cube was shown in [2].

The crossed cube CQn is a hypercube [15] variant with many attractive properties [1], [5], [4], [3], [7], [8], [9], [10], [11], [12], [19], [20]. Its diameter, wide diameter, and fault diameter values are approximately half the values of the hypercube [1]. The crossed cube can embed binary trees and its cycle is slightly more than the hypercube [1], [9], [12], [19]. The various path lengths have been embedded in crossed cube as shown in [8]. Under the comparison diagnosis model, the fault diagnostic ability of crossed cube is the same as that of the hypercube [7]. In [10], the authors provided fault-tolerant Hamiltonian path embedding in a crossed cube. Yang et al. [18] further presented fault-tolerant cycle embedding in a crossed cube. Many-to-many disjoint path covers in crossed cube with faulty nodes and/or edges have been shown in [16]. In this paper, we show that a crossed cube can tolerate up to 2n3 faulty nodes without being disrupted when each node exists at least one fault-free neighbor. Furthermore, the conditional fault diameter of a crossed cube is shown to be (n+1)/2+3 with 2n3 faulty nodes. Therefore, the conditional fault diameter equals three plus the diameter. With this result, the conditional fault diameter of a crossed cube is also approximately one-half that of the hypercube.

The rest of this paper is organized as follows. Section 2 summarizes results on crossed cubes and gives the necessary background and notations used in this paper. In Section 3, we prove the conditional connectivity and conditional fault diameter of crossed cubes. Section 4 presents our conclusions.

Section snippets

Preliminaries and notations

Let G=(V,E) represent a graph in which V represents the node set of G and E is the edge set of G. The topology of an interconnection network is usually denoted by a graph G=(V,E), where every node vV represents a processor and every edge (u,v)E represents a link between u and v. Let x and y be two nodes. The distance between x and y in G is d(x,y) and D(G) represents the diameter of G. The set N(v) denotes all of the neighbors of v. Let S be a set of nodes. The set S is restricted to the set

Conditional connectivity and conditional fault diameter of crossed cubes

Let v be a node of CQn and let F={XV| for any node vV,N(v)X} be an arbitrarily conditional faulty set. Note that the conditional faulty set represents each node with at least one fault-free neighbor. The conditional connectivity κF(CQn) of the crossed cube CQn is defined as the minimum cardinality |F| of nodes such that the subgraph CQnF is disconnected. The conditional connectivity of CQn is proven in the following lemmas.

Lemma 3

κF(CQn)2n2 for all n2 .

Proof

Consider the arbitrary edge (u,v) in CQn.

Conclusions

The conditional connectivity provides accurate measures of fault tolerance for a network. Moreover, conditional fault diameter represents more realistic delay time in communication when node faults occur. We have established the conditional connectivity and conditional fault diameter of crossed cubes CQn under the conditional faulty sets, which means that each non-faulty node has at least one non-faulty neighbor. Under this condition, the conditional connectivity of CQn is 2n2 for n2.

Chien-Ping Chang received the B.S. degree in Electrical Engineering from Chung Cheng Institute of Technology in 1986, and the Ph.D. degree in Computer and Information Science from National Chiao Tung University, Taiwan, Republic of China, in 1998. He is currently an associate professor in the Department of Electrical and Electronic, Chung Cheng Institute of Technology, Taiwan, Republic of China. His research interests include parallel computing, interconnection networks, graph theory, image

References (20)

  • K. Efe et al.

    Topological properties of the crossed cube architecture

    Parallel Computing

    (1994)
  • P. Kulasinghe

    Connectivity of the crossed cube

    Information Processing Letters

    (1997)
  • Y. Rouskov et al.

    Conditional fault diameter of star graph networks

    Journal of Parallel and Distributed Computing

    (1996)
  • M.C. Yang et al.

    Fault-tolerant cycle-embedding of crossed cubes

    Information Processing Letters

    (2003)
  • C.P. Chang et al.

    Edge congestion and topological properties of crossed cubes

    IEEE Transactions on Parallel and Distributed Systems

    (2000)
  • K. Day

    The conditional node connectivity of K-ary n-cube

    Journal of Interconnection Networks

    (2004)
  • K. Efe

    A variation on the hypercube with lower diameter

    IEEE Transactions on Computers

    (1991)
  • K. Efe

    The crossed cube architecture for parallel computing

    IEEE Transactions on Parallel and Distributed Systems

    (1992)
  • A.H. Esfahanian

    Generalized measures of fault tolerance with applications to n-cubes networks

    IEEE Transactions on Computers

    (1989)
  • J. Fan

    Diagnosability of crossed cubes under the comparison diagnosis model

    IEEE Transactions on Parallel and Distributed Systems

    (2002)
There are more references available in the full text version of this article.

Cited by (12)

  • Topological properties on the diameters of the integer simplex

    2015, Discrete Applied Mathematics
    Citation Excerpt :

    The diameters of Cartesian product graphs were studied in [1–3,6,7,17]. The parameters of some well-known networks such as hypercube, crossed cube, etc., were studied in [4,5,9,11,12,15,18]. The rest of this paper is organized as follows.

  • Embedding meshes/tori in faulty crossed cubes

    2010, Information Processing Letters
  • Fault diameter of hypercubes with hybrid node and link faults

    2009, Journal of Interconnection Networks
  • Embedded connectivity of some BC networks

    2022, Journal of Supercomputing
  • The conditional connectivity and restricted connectivity of ECQ(s,t)

    2019, International Journal of Circuits, Systems and Signal Processing
  • Reliability measures of crossed cube networks

    2018, International Conference on Information Networking
View all citing articles on Scopus

Chien-Ping Chang received the B.S. degree in Electrical Engineering from Chung Cheng Institute of Technology in 1986, and the Ph.D. degree in Computer and Information Science from National Chiao Tung University, Taiwan, Republic of China, in 1998. He is currently an associate professor in the Department of Electrical and Electronic, Chung Cheng Institute of Technology, Taiwan, Republic of China. His research interests include parallel computing, interconnection networks, graph theory, image processing, and data hiding.

Chia-Ching Wu received the B.S. degree in Electrical Engineering from Chung Cheng Institute of Technology in 1993, and the M.S. degree in Computer Science and Information Engineering from National Taiwan University in 2003. Currently, he is a Ph.D. candidate in Chung Cheng Institute of Technology and his research interests include parallel computing, interconnection networks, graph algorithms.

This work was supported in part by the National Science Council of Republic of China under Contract NSC 92-2213-E-014-010.

View full text