Trap array

Abstract

Scalable routing for large-scale wireless networks needs to find near shortest paths with low state on each node, preferably sub-linear with the network size. Two approaches are considered promising toward this goal: compact routing and geometric routing (geo-routing). To date the two lines of research have been largely independent, perhaps because of the distinct principles they follow. In particular, it remains unclear how they compare with each other in the worst case, despite extensive experimental results showing the superiority of one or another in particular cases. We develop a novel Trap Array topology model that provides a unified framework to uncover the limiting behavior of ten representative geo-routing algorithms. We present a series of new theoretical results, in comparison with the performance of compact routing as a baseline. In light of their pros and cons, we further design a Compact Geometric Routing (CGR) algorithm that attempts to leverage the benefits of both approaches. Theoretic analysis and simulations show the advantages of the topology model and the algorithm.