A nearly optimal upper bound for the self-stabilization time in Herman’s algorithm
Self-stabilization algorithms are very important in designing fault-tolerant distributed systems. In this paper we consider Herman’s self-stabilization algorithm and study its expected termination time. McIver and Morgan have conjectured the optimal upper bound being \(0.148N^2\), where \(N\) denotes the number of processors. We present an elementary proof showing a bound of \(0.167N^2\), a sharp improvement compared with the best known bound \(0.521N^2\). Our proof is inspired by McIver and Morgan’s approach: we find a nearly optimal closed form of the expected stabilization time for any initial configuration, and apply the Lagrange multipliers method to give an upper bound.
KeywordsHerman’s algorithm Self-stabilization Lagrange multipliers method
The authors would like to thank the anonymous reviewers for their insightful and helpful comments. This work was partially supported by the Australian Research Council (Grant No. DP130102764), the National Natural Science Foundation of China (Grant Nos. 61428208, 61472473, 61472412 and 61361136002), and the CAS/SAFEA International Partnership Program for Creative Research Teams.
- 4.Feller, W.: An Introduction to Probability Theory and Its Applications, vol. 1. Wiley, New York (1968)Google Scholar
- 5.Feng, Y., Zhang, L.: A tighter bound for the self-stabilization time in Herman’s algorithm. Inf. Process. Lett. 113(13), 486–488 (2013)Google Scholar
- 6.Feng, Y., Zhang, L.: A nearly optimal upper bound for the self-stabilization time in Herman’s algorithm. In: CONCUR, vol. 8704, pp. 342–356. Springer, Berlin (2014)Google Scholar