Range-Only Localization in n-Dimensional Networks With Arbitrary Anchor Placement

Abstract

This paper considers node localization in static sensor networks using range-only measurements. Similar to state- of-the-art algorithms, such as ECHO and DILOC, we rely on barycentric coordinates of the nodes to transform the non-convex node localization problem into a linear system of equations. The main contribution of this paper is a simple closed-form expression for generalized barycentric coordinates, which extends existing algorithms from two to n dimensions and allows arbitrary anchor-node configurations. The result relies on a connection between the Cayley-Menger bi-determinants of subsets of n+1 neighbor nodes and the signed volume of the simplices defined by these neighbor nodes. Hence, for noise-free measurements, the proposed method computes the optimal sensor network embedding as the solution of a linear system with coefficients obtained from the generalized barycentric node coordinates. Using simulations, we provide comparisons with DILOC and Matlab's MDS implementation. We also show that it is possible to improve our algorithm run time using fewer subsets of neighbor nodes.