Subcube reliability of a modular fault-tolerant hypercube architecture

The importance of hypercube networks motivated several researchers to propose, analyze and design fault-tolerant hypercube architectures. In this paper a modular fault-tolerant hypercube architecture using Fault Tolerant Basic Blocks (FTBBs) which can tolerate both processor and link failures is proposed. A reconfiguration strategy to tolerate processor and link failures within the FTBB is then given. In any Hypercube network it is important to know the probability of embedding the largest fault-free subcube (Subcube Reliability) in the presence of node and/or link failures. This is a measure of how efficiently hypercube algorithms can run on different fault-tolerant hypercubes. We investigate subcube reliability of the proposed fault-tolerant hypercube architecture for two failure models: the node failure model and the supernode failure model. The analysis shows that the fault-tolerant hypercube architecture we propose is more resilient than the basic hypercube in terms of its ability to support several smaller subcubes in the damaged structure.