2003 · 81 citations · 3 references
ComputingComputer EngineeringComputer ScienceAlgorithmsFault ToleranceFault-tolerant ComputingFault Tolerance In Parallel SystemsHardware Fault ToleranceHypercube NetworksNetwork Fault ToleranceFault Tolerant AlgorithmsGraph Embedding In Hypercube NetworksHamiltonian CyclesHamiltonian Cycles In HypercubesHypercube Network Fault ToleranceHypercube Topology
The authors show that in an n-dimensional hypercube (Q/sub n/), up to n-2 links can fail before destroying all available Hamiltonian cycles. They present an efficient algorithm which identifies a characterization of a Hamiltonian cycle in Q/sub n/, with as many as n-2 faulty links, in O(n/sup 2/) time. Generating a fault-free Hamiltonian cycle from this characterization can be easily done in linear time. An important application of this work is in optimal simulation of ring-based multiprocessors or multicomputer systems by hypercubes. Compared with the existing fault-tolerant embeddings based on link-disjoint Hamiltonian cycles, the algorithm specifies such a cycle that tolerates twice as many faulty links.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
3