Topology-sensitive epidemic algorithm for information spreading in large-scale systems

Abstract

Epidemic algorithms are an emerging technique that has recently gained popularity as a potentially effective solution for disseminating information in large-scale network systems. For some application scenarios, efficient and reliable data dissemination to all or a group of nodes in the network is necessary to provide with the communication services within the system. These studies may have a large impact in communication networks where epidemic-like protocols become a practice for message delivery, collaborative peer-to-peer applications, distributed database systems, routing in Mobile Ad Hoc networks, etc. In this paper we present, through various simulations, that an epidemic spreading process can be highly influenced by the network topology. We also provide a comparative performance analysis of some global parameters performance such as network diameter and degree of connectivity. Based on this analysis, we propose a new epidemic strategy that takes into account the topological structure in the network. The results show that the proposed epidemic algorithm outperform a classical timestamped anti-entropy epidemic algorithm in terms of the number of sessions required to reach a consistent state in the network system.