Active stabilization

Abstract

We propose the notion of active stabilization for computing systems. Unlike typical stabilizing programs (called passive stabilizing in this paper) that require that the faults are absent for a long enough time for the system to recover to legitimate states, active stabilizing programs ensure recovery in spite of constant perturbation during the recovery process by an adversary. We identify the relation between active and passive stabilization in terms of their behavior and by comparing their cost of verification. We propose a method for designing active stabilizing programs by a collection of passive stabilizing programs. Finally, we compare active stabilization with fault-contained stabilization and stabilization in the presence of Byzantine faults. © 2011 Springer-Verlag.