Fault-Tolerant Communications in Cube-Based Multiple-Bus Systems

We consider routing in multiple-bus systems with faulty nodes and buses. A cube-based multiple-bus system is used as an example to demonstrate our approach. In such a system, processors are partitioned into two sets based on the bipartite graph definition, one set is called a processor set and the other one becomes a bus set. A fault-tolerant routing scheme is proposed based on limited global information of fault distribution captured by an integer called safety level associated with each processor and bus. A safety level is an approximated measure of the number and distribution of faulty components in the neighborhood. Each processor (also called node) is considered as an active entity while each bus is viewed as a passive one. The safety level of each bus is calculated by one of its adjacent nodes. Optimal routing between two nodes is guaranteed if the safety level of the source node is no less than the Hamming distance between the source and destination nodes. The proposed routing scheme can also be used in disconnected cube-based multiple-bus systems, where nodes are disjointed. The feasibility of optimal or suboptimal routing can by easily determined at the source node by comparing its safety level, its neighbors’ safety levels, together with the Hamming distance between the source and destination nodes. The proposed scheme is the first attempt to address fault-tolerant communication in a multiple-bus system.