Anchor Selection for Localization in Large Indoor Venues

Abstract

Many indoor localization systems rely on a set of reference anchors with known positions. A target's location is estimated from a set of distances between the target and its surrounding anchors, and hence the selection of anchors affects the localization accuracy. However, it remains a challenge to select the best set of anchors. In this paper, we study how to appropriately make use of the surrounding anchors for localizing a target. We first construct different candidate anchor clusters by selecting different number of anchors with the strongest received signals. Then for each candidate cluster, we propose a weighted min-max algorithm to provide a location estimation. Finally, we introduce a weighted geometric dilution of precision (w-GDOP) algorithm that combines the estimations from multiple clusters by quantifying their estimation accuracy. We evaluate the performance of our solution through simulations and real-world experiments. Our results show that the proposed anchor selection scheme and localization algorithm significantly improve the localization accuracy in large indoor environments.