A statistical approach to localize passive RFIDs

We suggest a novel method for localizing passive radio frequency identification (RFID) tags based on a Bayesian approach. The localization framework requires a suitable number of tag-readers deployed in the area under investigation and connected to antennas which, provided with rotation ability, perform the detection task by scanning each angular sector at different power levels. Since readers are programmed to detect tags at increasing transmitting power (and provide binary tag/no tag information), we take advantage of the reader interrogation results to develop a Bayesian model for tags localization. By testing the solution with off-the-shelf UHF tags were experimented the accuracy of the method that, in a 5mtimes4m environment, provides an average error in localization of about 0.6 meters. This without exploiting any a priori information of the location, orientation or the power delivered to the tag (since not available in actual tags) which would significantly improve localization accuracy