Randomization Adaptive Self-stabilization
Self-stabilizing algorithms are designed to start from an arbitrary state and eventually exhibit a desired behavior. Self-stabilizing algorithms that use randomization are able to achieve tasks that cannot be achieved by deterministic means. In addition, in some cases, randomization enables faster convergence of self-stabilizing algorithms. Often, randomized self-stabilizing algorithms are designed to use an infinite amount of random bits to operate correctly. However, the creation of (real) random bits is considered expensive; thus, a randomization adaptive self-stabilizing algorithm which uses random bits during convergence but does not use random bits following the convergence is desirable. Such a notion of adaptiveness has been studied in the past, where the resource demands of a self-stabilizing algorithm are reduced upon convergence, be it memory requirements, or communication requirements.
- 1.Dolev, S., Tzachar, N.: Randomization Adaptive Self-Stabilization. CoRR, abs/0810.4440 (2008)Google Scholar