Distributed fault-tolerant routing on hypercubes algorithms and performance study

Increasing use of hypercube systems in reliability-critical applications has made fault-tolerant communication algorithms for hypercube important. This paper describes four fault-tolerant routing algorithms for hypercubes subject to link failures, namely any/sub -/nd, sidewalk, lookahead and lookside. The principle of sidewalk and lookahead are similar to two existing approaches. Lookside is an improved version of either of them. Sidewalk, lookahead and lookside guarantee successful routing in a d-cube if the number of link failures is less than d. For higher degree of link failures, the authors measure their performance by probability of successful routing and expected routing distance using two fault distribution models, probability fault model and random fault model. Lookside demonstrates better properties than sidewalk and lookahead in that it has the highest successful routing rate with reasonable routing distance.<
>