2-D localization of passive UHF RFID tags using location fingerprinting

Localization of UHF RFID tags in an industrial environments is difficult due to signal reflections and multipaths caused by steel and metal objects. Existing solutions have shown decent accuracy for small distances but fail to maintain the accuracy as the distance between the antenna and the tag increases. In this paper, we describe a novel UHF RFID localization approach based on location fingerprinting. The approach uses machine learning to transform localization into a classification problem. Location fingerprints are generated using outputs Bartlett beamformer and MUSIC algorithms that estimate the incoming angle of a signal. We evaluated our approach in an industrial environment, and the results show that we achieve a high classification accuracy and maintain it with the increase of the distance between the tag and the antenna.