Brief Announcement: Self-stabilizing Synchronization of Arbitrary Digraphs in Presence of Faults

Abstract

This brief announcement presents a fault-tolerant self-stabilizing distributed clock synchronization protocol for an arbitrary, non-partitioned digraph. Synchronization algorithms are essential for managing the use of resources and controlling communication in a distributed system. Synchronization of a distributed system is the process of achieving and maintaining a bounded skew among independent local time clocks. A distributed system is said to be self-stabilizing if, from an arbitrary state, it is guaranteed to reach a legitimate state in a finite amount of time and remain in a legitimate state. For clock synchronization, a legitimate state is a state where all parts in the system are in synchrony. The self-stabilizing distributed-system clock synchronization problem is, therefore, to develop an algorithm (i.e., a protocol) to achieve and maintain synchrony of local clocks in a distributed system after experiencing system-wide disruptions in the presence of network element imperfections. The convergence and closure properties address achieving and maintaining networksynchrony, respectively.