Fault-Tolerant Spanners: Better and Simpler

TL;DR

This paper introduces improved methods for constructing fault-tolerant graph spanners that are simpler, more efficient, and have better dependence on the number of faults, with applications to distributed algorithms.

Contribution

It presents a new transformation for fault-tolerant spanners, improves approximation ratios for specific cases, and develops distributed algorithms demonstrating the local nature of the constructions.

Findings

New transformation reduces edges in fault-tolerant spanners.

Improved approximation ratio to O(log n) for k=2 case.

Distributed algorithms show constructions are inherently local.

Abstract

A natural requirement of many distributed structures is fault-tolerance: after some failures, whatever remains from the structure should still be effective for whatever remains from the network. In this paper we examine spanners of general graphs that are tolerant to vertex failures, and significantly improve their dependence on the number of faults $r$, for all stretch bounds. For stretch $k \geq 3$ we design a simple transformation that converts every $k$-spanner construction with at most $f(n)$ edges into an $r$-fault-tolerant $k$-spanner construction with at most $O(r^3 \log n) \cdot f(2n/r)$ edges. Applying this to standard greedy spanner constructions gives $r$-fault tolerant $k$-spanners with $\tilde O(r^{2} n^{1+\frac{2}{k+1}})$ edges. The previous construction by Chechik, Langberg, Peleg, and Roddity [STOC 2009] depends similarly on $n$ but exponentially on $r$ (approximately like $k^r$). For the case $k=2$ and unit-length edges, an $O(r \log n)$-approximation algorithm is known from recent work of Dinitz and Krauthgamer [arXiv 2010], where several spanner results are obtained using a common approach of rounding a natural flow-based linear programming relaxation. Here we use a different (stronger) LP relaxation and improve the approximation ratio to $O(\log n)$, which is, notably, independent of the number of faults $r$. We further strengthen this bound in terms of the maximum degree by using the \Lovasz Local Lemma. Finally, we show that most of our constructions are inherently local by designing equivalent distributed algorithms in the LOCAL model of distributed computation.