Optimal Spanners for Unit Ball Graphs in Doubling Metrics

TL;DR

This paper proves the existence of light-weight, bounded-degree spanners for unit ball graphs in doubling metrics, introduces an efficient distributed algorithm in the LOCAL and CONGEST models, and provides experimental validation.

Contribution

It presents the first construction of light-weight, bounded-degree spanners for unit ball graphs in doubling metrics with efficient distributed algorithms and experimental validation.

Findings

Improved lightness bounds for spanners in doubling metrics.

Distributed algorithm with $ ext{O}( ext{log}^* n)$ rounds in LOCAL and CONGEST models.

Experimental results confirming theoretical bounds and efficiency.

Abstract

Resolving an open question from 2006, we prove the existence of light-weight bounded-degree spanners for unit ball graphs in the metrics of bounded doubling dimension, and we design a simple $\mathcal{O}(\log^*n)$-round distributed algorithm in the LOCAL model of computation, that given a unit ball graph $G$ with $n$ vertices and a positive constant $\epsilon < 1$ finds a $(1+\epsilon)$-spanner with constant bounds on its maximum degree and its lightness using only 2-hop neighborhood information. This immediately improves the best prior lightness bound, the algorithm of Damian, Pandit, and Pemmaraju, which runs in $\mathcal{O}(\log^*n)$ rounds in the LOCAL model, but has a $\mathcal{O}(\log \Delta)$ bound on its lightness, where $\Delta$ is the ratio of the length of the longest edge to the length of the shortest edge in the unit ball graph. Next, we adjust our algorithm to work in the CONGEST model, without changing its round complexity, hence proposing the first spanner construction for unit ball graphs in the CONGEST model of computation. We further study the problem in the two dimensional Euclidean plane and we provide a construction with similar properties that has a constant average number of edge intersections per node. Lastly, we provide experimental results that confirm our theoretical bounds, and show an efficient performance from our distributed algorithm compared to the best known centralized construction.