This paper presents an algorithm for embedding a ring of a given size in a damaged n-cube. Existing algorithms all attempt to find a ring which is as large as possible and can only tolerate up to 2n - ŁK() faulty nodes. There are two contributions made in this paper. First, we have identified a problem which is closer to a realistic situation in mapping a parallel algorithm to a hypercube machine. Second, the proposed algorithm can tolerate a significantly larger number (ŁK(2ⁿ/2)) of faulty nodes, and always embeds a ring with congestion 1 and dilation of at most 2. This result compares favorably to existing algorithms in embedding congestion, embedding dilation, or degrees of fault tolerance.

Keywords: graph embedding, hypercube, ring, linear path, fault tolerance

Received June 1, 1995; revised April 23, 1996.
Communicated by Wei-Pang Yang.
^*This work is supported in part by the National Science Council, R.O.C., under grant number NSC85-2213-E-216-021.