APDV: making distance vector routing scale using adaptive publish-subscribe mechanisms

Abstract

The Adaptive Publish-subscribe Distance Vector (APDV) protocol is introduced as an example of a new approach to allowing distance-vector routing to scale by integrating it with adaptive publish-subscribe mechanisms. APDV combines establishing routes to well-known controllers using distance-vector signaling with publish-subscribe mechanisms. The latter allow destinations to publish their presence with subsets of controllers, and sources to obtain routes to intended destinations from those same controllers. Controllers are selected dynamically using a fault-tolerant distributed election algorithm to ensure that each non-controller node is covered by at least a given number of controllers within a few hops. Extensive simulation experiments are used to compare APDV with AODV and OLSR, which are representative protocols for on-demand and proactive routing. The results show that APDV achieves significantly better data delivery, attains comparable delays for delivered packets, and incurs orders of magnitude less control overhead than AODV and OLSR, even under heavy data loads.