Towards a characterization of localization performance in networks with random geometries

Abstract

This paper introduces a paradigm shift in the way localization performance in wireless networks is analyzed. Currently, localization performance in networks is commonly benchmarked using the Cramer-Rao lower bound (CRLB), given a fixed geometry of anchor nodes and a target. By contrast, we consider networks with random geometries for which the CRLB itself becomes a random variable, and we seek a tractable expression for its distribution. The ultimate goal is to derive a complete statistical characterization of localization performance throughout an entire network via analysis, rather than resorting to simulation. This paper presents an initial step towards achieving this goal by using an approximation of the CRLB based on angles between anchor nodes. This facilitates the derivation of an accurate approximation of the distribution of the CRLB, conditioned on the number of hearable anchors.