Optimization of an RFID location identification scheme based on the neural network.

An indoor localization technology is increasingly critical as location-aware applications evolve. Researchers have proposed several indoor localization technologies. Because most of the proposed indoor localization technologies simply involve using the received signal strength indicator value of radio-frequency identification (RFID) for indoor localization, radio-frequency interference, and environmental factors often limit the accuracy of localization results. Therefore, this study proposes an accurate RFID localization based on the neural network (ARL-N2), a passive RFID indoor localization scheme for identifying tag positions in a room, combining a location identification based on dynamic active RFID calibration algorithm with a backpropagation neural network (BPN). The proposed scheme composed of two phases: in the training phase, an appropriate BPN architecture is constructed using the training data derived from the coordinates of reference tags and the coordinates obtained using the localization algorithm. By contrast, the online phase involves calculating the tracking tag coordinates and using these values as BPN inputs, thereby enhancing the estimated location. A performance evaluation of the ARL-N2 schemes confirms its high localization accuracy. The proposed method can be used to locate critical objects in difficult-to-find areas by creating minimal errors and applying and economical technique. Copyright © 2013 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]