Análisis de contención de fallas en algoritmos auto-estabilizantes de elección de líder
"Self-stabilization is a relatively new paradigm for fault tolerance in distributed systems. The goal of self-stabilization is to maintain a system free of errors. A selfstabilizing algorithm helps a system to recover from transient faults in a finite number of steps, without involving any exte...
| Autor: | |
|---|---|
| Formato: | tesis de maestría |
| Estado: | Versión publicada |
| Fecha de publicación: | 2010 |
| País: | México |
| Recursos: | Centro de Investigación Científica y de Educación Superior de Ensenada |
| Repositorio: | Repositorio Institucional CICESE |
| Idioma: | español |
| OAI Identifier: | oai:cicese.repositorioinstitucional.mx:1007/1892 |
| Acesso em linha: | http://cicese.repositorioinstitucional.mx/jspui/handle/1007/1892 |
| Access Level: | acceso abierto |
| Palavra-chave: | info:eu-repo/classification/Autor/Auto-estabilización info:eu-repo/classification/Autor/Self-stabilization info:eu-repo/classification/Autor/Elección de líder info:eu-repo/classification/Autor/Leader election info:eu-repo/classification/Autor/Fallas info:eu-repo/classification/Autor/Faults info:eu-repo/classification/Autor/Contención info:eu-repo/classification/Autor/Containment info:eu-repo/classification/cti/1 info:eu-repo/classification/cti/11 |
| Resumo: | "Self-stabilization is a relatively new paradigm for fault tolerance in distributed systems. The goal of self-stabilization is to maintain a system free of errors. A selfstabilizing algorithm helps a system to recover from transient faults in a finite number of steps, without involving any external agent. There are many graph problems that have been solved by self-stabilizing algorithms. This thesis deals with one of them, the leader election problem, whose definition is as follows. Given a graph G and set of nodes C of G, the idea is to elect a unique node from the set C. A fault containment technique helps self-stabilizing algorithms to recover faster from transient faults. This technique avoids that the effect of a fault propagates to healthy nodes of the graph. The objective of this thesis is to incorporate a fault containment technique to a self-stabilizing leader election algorithm. Our algorithm is restricted to run on anonymous trees. We proved the correctness of the algorithm and show that it terminates in O(n4) time, starting in any arbitrary state. Once the tree is in a legal state and a simple fault occurs, the proposed algorithm takes the tree to a legal state in constant time. The most significant aspect of the algorithm is that the fault –gap is independent of the network size." |
|---|