Radio frequency identification (RFID) is a promising technology for localization in various industrial applications. In RFID localization, accuracy is the top performance concern, and it is affected by multiple factors. In this paper, we investigate how the facility geometry impacts the expected localization accuracy in the entire region where the target is uniformly distributed. Three groups of geometries, namely, rectangles with various length-to-width ratios, circle, and regular polygons with 3–10 edges, are chosen for this study. A hybrid multilateration approach, which combines linearization and nonlinear optimization, is used to estimate the target location. Since the layout of landmarks significantly affects localization performance, we evaluate the expected accuracy in a facility obtained under the optimal landmark layout for the facility. The optimal landmark layout for each type of facility geometry is obtained, and then the effect of geometry is studied by comparing the expected accuracies of these layouts. It is discovered that (1) the optimal layouts follow several simple empirical deployment principles, (2) for all geometries, the expected accuracy improves and tends to reach the expected Cramer-Rao lower bound as more landmarks are used, and (3) if the same numbers of landmarks are used, the expected accuracies for circular and regular polygonal geometries are close. However, the expected accuracy for a rectangular geometry decreases as the length-to-width ratio increases.

This content is only available via PDF.
You do not currently have access to this content.