Epidemic-style Communication Primitives Exploiting Network Topology Information

Gossip-based communication and epidemic-style routing algorithms have been proposed to achieve scalability in distributed systems and to support probabilistic communication when the application of the classic deterministic algorithms and protocols is impossible or unsuitable. In this paper, we present a middleware for probabilistic communication that relies on optimised epidemic-style information dissemination techniques for distributed systems, based on recent results of complex networks theory. The novelty of our approach resides in the evaluation and the exploitation of the structure of the underlying network for the automatic tuning of the dissemination process to control. With respect to unicast communication, we show that protocols that statistically ensure the desired reliability level in the case of homogeneous networks (i.e., a network composed of nodes with the same degree of connectivity) can be designed. We demonstrate that these results can be exploited also in the case of anycast and multicast communication to tune and optimise the replication process. We evaluate our approach analysing the application of these techniques to the case of mobile ad hoc systems, a specific class of distributed systems. Finally, we generalise the model to the case of heterogeneous networks.