Distributed topology discovery in resource-constrained IoT: a Directed Flooding approach

As IoT networks become larger and more diversified, service discovery becomes a cumbersome task. Nodes need to be polled and the upkeep of routing tables might not be possible in low-power devices as this comes with a considerable energy cost and added complexity. For these reasons, flooding is a very well-known strategy for building large peer networks with low-energy protocols. But flooding requires blindly spreading information until the destinations have been reached and is poorly suited to unicast traffic. In this paper we explore a fully distributed topology discovery in low-energy networks, we show that flooding can be a great strategy for early communication while a topology is built. We investigate which additional information should be added to a data packet in order to build the topology quickly and reliably and what is the trade-off in terms of communication overhead. Furthermore, we show that not all nodes need to be queried for a reliable topology discovery but traditional centrality measures can be exploited and specific nodes can be targeted based on the partial topology each node is able to build over time.