Fault tolerance of metric basis can be expensive

TL;DR

This paper investigates the cost of achieving fault tolerance in metric bases of graphs, demonstrating that the required number of sensors can grow exponentially relative to the metric dimension, and constructs graphs to prove this bound is tight.

Contribution

The paper constructs graphs showing the exponential necessity of additional sensors for fault-tolerant metric bases and extends these results to the k-metric dimension.

Findings

Exponential upper bounds on fault-tolerant metric dimension are necessary.

Constructed graphs demonstrate the tightness of exponential bounds.

Extended results to the k-metric dimension, showing similar exponential growth.

Abstract

A set of vertices S is a resolving set of a graph G; if for every pair of vertices x and y in G, there exists a vertex s in S such that x and y differ in distance to s. A smallest resolving set of G is called a metric basis. The metric dimension dim(G) is the cardinality of a metric basis of G. The notion of a metric basis is applied to the problem of placing sensors in a network, where the problem of sensor faults can arise. The fault-tolerant metric dimension ftdim(G) is the cardinality of a smallest resolving set S such that S\{s} remains a resolving set of G for every s in S. A natural question is how much more sensors need to be used to achieve a fault-tolerant metric basis. It is known in literature that there exists an upper bound on ftdim(G) which is exponential in terms of dim(G); i.e. ftdim(G) <= dim(G)(1+2^(5dim(G)-1)). In this paper, we construct graphs G with ftdim(G) = dim(G)+2^(dim(G)-1) for any value of dim(G), so the exponential upper bound is necessary. We also extend these results to the k-metric dimension which is a generalization of the fault-tolerant metric dimension. First, we establish a similar exponential upper bound on dim(k+1)(G) in terms of dim(k)(G); and then we show that there exists a graph for which dim(k+1)(G) is indeed exponential. For a possible further work, we leave the gap between the bounds to be reduced.