Active RFID Indoor Positioning and Navigation Based on Probability Method

Although RFID-based indoor positioning and navigation systems have been developed by many researchers over the world, improving positioning and navigation accuracy to a higher level (e.g., better than 1 m) is still a challenging task. Conventionally, the received signal strength is converted to a tag-reader range and the location is determined using trilateration. However, in a typical indoor environment, it is difficult to model precisely the relationship of received signal strength and a tag-reader range due to multipath. In this paper, we proposed a probabilistic localization approach to handle uncertainties and errors in the received signal strength. A signal-strength-to-range observation model was built based on the probabilistic characteristics of sample calibration data. To deal with nonlinear, non-Gaussian and non-stationary problems in indoor positioning and navigation, a redistributed particle filter was developed. Although our research and development is still in an early stage, we could attain promising results by improving the observation model and sub-optimal estimator used for the particle filter.