An UHF RFID Indoor Localization System based on the Maximum Likelihood Estimator with an Effective Tags Deployment Density and a Sub-Metric Accuracy

RFID-based indoor localization has received extensive attention from both industry and academia. Most of those systems use the Received Signal Strength (RSS) as positioning metric, thanks to its readability on the majority of user devices. RSS values are easily affected by the surrounding environment, that induces signals fluctuation due to multipath and to the propagation channel non-stationarity. To overcome the impact of these variations, we propose an original combining technique for the RSS samples, using the Maximum Likelihood Estimator (MLE). This approach improves both the fingerprinting reliability and the localization accuracy. Following on measurements in a classroom, our architecture features better results than conventional UHF RFID positioning systems relying on RSSIs averaging; it achieves a position error of 90 centimeters, with an RFID tag deployment density reduced to 0.062 tag per square meter.