Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Stabilization Anish Arora
 

Summary: Stabilization
Anish Arora
Department of Computer Science
The Ohio State University, Columbus
The concept of system stabilization has enjoyed growing attention from computer scientists over the past
25 years. Intuitively speaking, stabilization refers to the ability of a system to converge within nitely
|and possibly boundedly| many computation steps, from arbitrary system states to states from where
the system exhibits desired behavior.
Historically speaking, this concept was introduced to computer science by Edsger W. Dijkstra [1]. He coined
the term \self-stabilization", although our understanding is that he later concluded that the emphasis on
\self" was less than helpful. Since the concept characterizes just one of several interesting system properties,
the less anthropomorphic term \stabilization" will be used instead throughout this chapter.
The most frequently cited motivation for designing stabilization is to provide a sort of fault-tolerance,
whereby the system automatically recovers from fault occurrences by eventually resuming its desired be-
havior [2]. The faults considered most often along these lines are \transients", which corrupt the system
state (be it control or data) in an arbitrary manner. Empirical evidence exists that transient faults are
common in practice and, since it is not always possible for the system to exhibit desired behavior if they
occur, the alternative that the system eventually exhibits desired behavior is well motivated. That said,
it is important to point out that stabilization can be considered for other types of faults, e.g. process
crashes and restarts, Byzantine failures, etc. If, for instance, process crashes and restarts are coupled with

  

Source: Arora, Anish - Department of Computer Science and Engineering, Ohio State University

 

Collections: Computer Technologies and Information Sciences