Local Topology Aware Probabilistic Routing

Abstract

Gossip-based packet forwarding is used in unstructured networks is to reduce traffic overhead in dense networks and to minimize early gossip termination in sparse networks. Unlike in flooding, where packets are forwarded to all the neighbors, in Gossip-based protocols packets are forwarded with some probability value p<1, to reduce redundancy. However this value has to be carefully tuned: if too small, early gossip termination is likely to occur, if too large, flooding storms can take place, as with the flooding protocol. In this work, we propose to use a forwarding probability based on local topology indicators, such as the effective node degree of the forwarding node: the choice of such probability takes into account the local topology. In a context where each node can have a different forwarding probability, another way of setting efficiently its value consists in further tuning such value for each message, based on the estimated level of completion of the corresponding communication task: to this purpose we propose to use a simple formula based on the messages hop-count. We validate these approaches by simulation using ns-2 in sparse and dense networks and show that they improve the performances in terms of traffic overhead and average end-to-end delay. In terms of packet delivery ratio, the proposed approach yields results comparable to those of the standard protocol AODV.