Decentralized Cluster Detection in Distributed Systems Based on Self-Organized Synchronization

Abstract

In this work, we propose a method for the decentralized detection of clusters, or communities, in large-scale networked systems. Different from other approaches that require global knowledge of the network topology, the proposed method is based on a fully decentralized protocol and allows a node to infer knowledge about the community memberships of its nearest neighbours. It relies on the fact that topological characteristics of a network leave traces in the evolution of a self-organized synchronization process. The preliminary results presented in this report show a promising detection accuracy and justify a further investigation of our approach.