Fault-tolerant self-routing algorithms for hypercube based architectures

A key aspect in developing general hardware and software distributed architectures concerns the routing of messages among several processors. The aim of such routing algorithms is to guarantee the message delivery through shortest paths and to present a certain tolerance in case of link or node failures. The paper presents two routing algorithms suitable for message routing in a general purpose distributed system based on a N cube architecture. The decision policy, adopted in both the algorithms, is based on a particular assignment of labels to each node and requires minimal information, local to each node or embedded in the messages. Their simplicity, moreover, permits us to have a reduced overhead; an aspect of remarkable importance in real time systems.