Title :
A Systolic Design for Connectivity Problems
Author_Institution :
Department of Computer Science, North Carolina State University, Ralcigh, NC 27650.
Abstract :
In this paper we present a design, suited to VLSI implementation, for a one-dimensional array to solve graph connectivity problems. The computational model is relatively primitive in that only the two end cells of the array can interact with the external environment and only adjacent cells in the array are allowed to communicate. However, we show that an array of n + 1 cells can be used for a graph with n vertices to find the connected components, a spanning tree, or, when used in conjunction with a systolic priority queue, a minimum spanning tree. The area, time, and I/O requirements compare favorably with other models proposed for this problem in the case of sparse graphs.
Keywords :
Computational modeling; Computer science; Data structures; Digital systems; Global communication; Processor scheduling; Systolic arrays; Testing; Tree graphs; Very large scale integration; Connectivity problems; VLSI algorithms; systolic arrays; transitive closure; union-find data structure;
Journal_Title :
Computers, IEEE Transactions on
DOI :
10.1109/TC.1984.5009318