Classes of self-stabilizing protocols

Abstract

Self-stabilization is an abstraction of fault tolerance for transient malfunctions. Intuitively, a self-stabilizing system is a system which can be started from any possible state. The authors here explore the possibility of transforming an arbitrary distributed protocol into a self-stabilizing one. It is proved that some conditions are sufficient for this transformation to be feasible. These conditions lead to a classification of distributed protocols according to the complexity for performing their transformation.<>