Time optimal self-stabilizing synchronization

Abstract

In the network synchronization model, each node maintains a local pulse counter such that the advance of the pulse numbers simulates the advance of a clock in a synchronous network. In this paper we present a tame optimai sel&stabilizing scheme for network synchronization.

Asynchronous and Fully Self-stabilizing Time-Adaptive Majority Consensus

“…The algorithm utilizes some known techniques, namely the self-stabilizing synchronization [6] and the power-supply method [1]. We use a new version of the power-supply method.…”

“…The study of self-stabilizing protocols was initiated by Dijkstra [14]. Reset-based approaches to self-stabilization are described in [27,3,7,6,16,5]. One of the main drawbacks of this approach is that the detection mechanism triggers a system-wide reset in the face of the slightest inconsistency.…”

“…We use the self stabilizing synchronizer with a counter (sometimes called phase clock) of [6]. Other phase clocks in the literature, such as [10,18,13,23], may also be useful.…”

Adaptive Stabilization of Reactive Protocols

“…In this approach, the idea is that a special stabilizing mechanism monitors the system for illegal states, and whenever an inconsistency is detected, it invokes a special stabilizing reset protocol that imposes some legal global state on the system. The best reset protocols offer stabilization with relatively low space and communication overhead [10,9,11], but at a price of inherently high stabilization time-the cross-network latency time. Still, the time complexity of reset is better than the time-out approach for the following reason.…”

“…(The correct state may either be some agreed upon fixed state, or a state that is in some sense "close" to the faulty state [16].) The best reset protocols in terms of time are given in [9], where the stabilization time is proportional to the diameter of the network. Logarithmic space protocols are given in [4,10], and a randomized constant-space protocol is given in [19].…”