Summary: SelfStabilizing Unidirectional Network Algorithms by
Yehuda Afek y
Anat Bremler z
Computer Science Department,
TelAviv University, Israel 69978.
Powersupply, a surprisingly simple and new general paradigm for the development of
selfstabilizing algorithms in different models, is introduced. The paradigm is exemplified
by developing simple and efficient selfstabilizing algorithms for leader election and either
BFS or DFS spanning tree constructions, in stronglyconnected unidirectional and bi
directional dynamic networks (synchronous and asynchronous). The different algorithms
stabilize in O(n) time in both synchronous and asynchronous networks without assuming
any knowledge about the network topology or size, where n is the total number of nodes.
Following the leader election algorithms we present a generic selfstabilizing spanning tree
and/or leader election algorithm that produces a whole spectrum of new and efficient
algorithms for these problems. Two variations that produce either a rooted Depth First
Search tree or a rooted Breadth First Search tree are presented.
A distributed system is selfstabilizing if never mind what local state each of its processors