Fast reconfigurable network for graph connectivity and transitive closure

Hussein M. Alnuweiri*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

This paper presents a VLSI array for labeling the connected components of a graph on N nodes in O(r) steps using a reconfigurable bus of width m bits, such that (rm)≥N and 1≤r≤m. The network architecture consists of an array of simple logic nodes which are connected by a reconfigurable bus.To solve a problem on N nodes, the array uses N processors and N(N-1)/2 switches. The proposed connectivity and transitive closure algorithms are based on a processor indexing scheme which employs constant-weight codes. It is shown that when r is a constant, then the algorithm takes O(1) time using a bus of width O(N1/r), and when r = [log N/loglog N], the algorithm takes O(log N/loglog N) time using a bus or width O(log N) bits.

Original languageEnglish
Pages (from-to)105-115
Number of pages11
JournalParallel Processing Letters
Volume4
Issue number1-2
Publication statusPublished - Jun 1994
Externally publishedYes

Fingerprint

Dive into the research topics of 'Fast reconfigurable network for graph connectivity and transitive closure'. Together they form a unique fingerprint.

Cite this