Self-stabilization in spite of frequent changes of networks: case study of mutual exclusion on dynamic rings

  • Authors:

  • Toshimitsu Masuzawa;Hirotsugu Kakugawa

  • Affiliations:

  • Osaka University, Toyonaka, Japan;Osaka University, Toyonaka, Japan

  • Venue:
  • SSS'05 Proceedings of the 7th international conference on Self-Stabilizing Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

It is generally said that a self-stabilizing protocol is inefficient in distributed systems with frequent faults or topological changes and, what is worse, it might never converge to its intended behavior forever. Its main reason is that a new fault or topological change brings the system into an unexpected configuration, and thus, the system restarts convergence to its intended behavior from scratch. But the reasoning seems too pessimistic. This paper provides a novel observation about self-stabilization on frequently changing networks: by quantifying influence of steps of a self-stabilizing protocol and that of a topological change, efficiency of the convergence can be estimated with considering topological changes that occur during the convergence. To show the feasibility and effectiveness of the approach, this paper presents a simple self-stabilizing mutual exclusion protocol on a dynamic ring where processes can join and leave the ring at any time. This paper clarifies what restrictions on frequency of joins and leaves are sufficient to guarantee the convergence and to guarantee the intended behavior after the convergence. The restrictions are not strict and thus the protocol can complete convergence and can continue its intended behavior on a frequently changing ring.