Distributed Adaptive Protocols for Information Dissemination in Large-Scale Communication Systems

One approach for information dissemination in large-scale communication systems is using epidemic protocols. Current epidemic protocols, however, adopt a constant fanout policy, which does not enable end users to control the information dissemination process. For distributed applications that need to compute a global function within a pre-determined response time, better procedures to control the information dissemination process have to be developed. In this paper, we introduce two distributed adaptive epidemic protocols using a dynamic fanout scheme. They are named round-based dynamic fanout (RBdf) and cluster-based dynamic fanout (CBdf). In RBdf, the network topology is flat and each node transmits a message with a varied fanout every round. In CBdf, the network topology is hierarchical, and the fanout values in every cluster differ within the same round. The main objectives are to ensure that peers receive messages within a bounded latency and that the system message overhead is a bounded value. The performance of the proposed protocols are verified through both theoretical and simulation studies.