Error reduction in distance estimation of RSS propagation models using Kalman filters

Abstract

In this paper, we propose an indoor localization system that integrates the received signal strength (RSS) correction methods with probabilistic propagation models. The proposed system aims to achieve accurate modeling of signals' propagation inside buildings without the need for expensive site surveys. This is achieved by eliminating the environmental noise causing temporal variations in the RSS measurements, using the Kalman filtering technique. In addition, we use Gaussian Process Regression (GPR) to model the signal propagation inside buildings as a function of distance. Our experimental results support our argument that GPR models outperform the conventional path-loss model. In addition, our results also show that integrating Kalman filters with GPR models improves the accuracy of distance estimation by almost 2 meters.