Performance of RSS-based localization in unknown environments

Abstract

Localization in unknown environments uses received signal strengths (RSSs) to precisely estimate the position of an object or a person. Existing work on RSS-based localization in unknown environments, however, employs theoretical models for the path loss, leading to a simple but unrealistic prediction of localization accuracy in an actual operating environment. In this paper, we perform experiments to collect RSSs and quantify performance of two prominent methods of localization— the general-trust-region method and the iterative method. Performance measure is a commonly-used distance error, which is the distance between the estimate position and the actual position of the target node. The experiments occur in a stadium with a sufficient number of anchors installed. We collect the power of wireless signals that are transmitted from each anchor and are received at various positions in the stadium. We input the RSS values into the two localization methods and obtain the resulting distance errors. The results show that, for most positions and at a significance level of 1%, the iterative method is statistically more accurate than the general-trust-region method. The average distance error of the iterative method is approximately 6 m, while that of the general-trust-region method is approximately 9 m. From the experimental data, we are able to recommend approaches to improve positional accuracy, including a suitable model for RSSs. The results in this paper have practical utility and reveal a realistic performance of localization methods in unknown environments.