Proactive versus reactive routing in low power and lossy networks: Performance analysis and scalability improvements

In this paper, the classical debate on suitability of proactive versus reactive routing approaches is revisited, however in the context of their application in Low-Power Lossy Networks (LLNs) as opposed to pure Mobile Ad Hoc Networks (MANETs). We argue that LLNs differ from traditional ad hoc networks not only due to node capacity, but also in the nature of traffic generation and routing requirements. This study aims at a fair comparison between two protocols proposed at the IETF, namely RPL (IPv6 Routing Protocol for Low-Power and Lossy Networks) – the proactive candidate, and LOADng (LLN On-demand Ad-hoc Distance vector routing protocol – next generation) – the reactive candidate, making use of real traffic scenarios and topology deployments of varied size, as well as random topologies in the particular case of metrics relating to scalability. Several metrics of interest are investigated, including metrics that have not been paid much attention in the existing MANET literature. In the course of this investigation, we also uncovered non-optimal protocol behavior for the case of large networks and proposed new mechanisms that are shown to improve control plane congestion, as a result improving network lifetime for large scale LLNs. We also derive bounds on control overhead for the two protocols, which indicate that RPL incurs an overhead that is lower than O(N2)O(N2) and reduces to Θ(Nlog(N))Θ(Nlog(N)) for a balanced tree structure, whereas LOADng has a total control overhead of Θ(N2)Θ(N2), irrespective of topology.