Fault tolerance in rearrangeable networks

Following a brief introduction to multistage interconnection networks in general and the Benes rearrangeable network in particular, the author presents techniques for achieving fault tolerance through appropriate manipulation of faulty networks without adding any hardware components. The author also shows how different types of faults at different locations can be tolerated with minimal performance degradation. Since a number of multistage blocking networks are topologically equivalent, the techniques described may be adapted to other rearrangeable network configurations, whether they are symmetrical or asymmetrical. Several examples are given to illustrate the techniques presented