Investigation of RFID tag motion estimation in real-world scenarios

Abstract

Radio Frequency Identification, already incorporated into several fields such as supply chain management, theft control and inventorying, still has a lot of potential to grow. It is possible to gain more information from tagged objects, than just the serial number. The signal response from an RFID tag present the RFID reader with much more than just a code that you can find on a barcode. Phase and RSSI are only two of the several parameters that form part of the signal response from an RFID tag. It is possible to use these additional signal response parameters to perform a motion estimation of a moving tag. Motion estimation is defined mainly as the velocity at which the tag is moving, the general direction of travel and a broad distance estimation. Building on previous work it is important to consider not only the abilities but also the limitations of these methods and how the techniques can be adjusted to adhere to all standards, in order to be implemented in real-world scenarios. Previous work in this field did not include all the local standards, placing limitations on the practical application of these techniques. Only limited characterization of real-world challenges, encountered when using these techniques, were conducted by previous research. Thus, a further study on the impact of these limitations as well as the influences from real-world environments is necessary. Against this background the aim for this study is to investigate the possibilities offered by motion estimation of RFID tags while adhering to local standards, and to characterize the effects from real-world conditions on the implementation of these methods. Thorough literature studies were performed on the following topics: challenges posed by RFID, localization mapping systems, direction estimation with RFID, phase difference on arrival techniques for velocity estimation, indoor location systems, Kalman filtering, phase unwrapping and antenna setups. Practical experimentation was conducted to determine the ideal antenna setup for motion estimation techniques. This was based on the literature review and was implemented with commercial off-the-shelf (COTS) equipment to inform further experimentation later in the study. Complete designs for the velocity estimation, direction estimation and distance estimation techniques were presented. These designs include practical limitations and implementation requirements using COTS equipment. Kalman filter designs were implemented for different scenarios using different combinations of measurements as inputs. These designs include the models used for motion estimation and the covariances for the different noise parameters present. A Kalman filter for sensor fusion was designed and implemented using both distance estimation from the signal strength parameter and phase-based velocity estimation to predict the motion of a moving tag. Experimental designs on how to test the different motion estimation techniques in real-world scenarios are discussed. The different influences on the accuracy of the techniques are investigated by means of simulated real-world conditions. Worst case scenarios are explored and possible applications in the case of inaccurate measurements are proposed. Comparisons between different velocity estimation techniques in a variety of different controlled circumstances were completed. These comparisons serve as a benchmark of how accurate the estimation technique is, as well as which estimation technique is better suited to what environments. From this study, it is also possible to determine which environmental conditions will influence the accuracy of these techniques and what the limits are of these techniques. The techniques were implemented to function both in a real time application as well as to record first and process later. Finally, it is proven that motion estimation of a RFID tag is possible in certain real-world scenarios, while adhering to local broadcasting channels. The practical benefit of this study is that more information can be provided by RFID systems, without additional hardware cost. This allows industrial and commercial users to improve the monitoring of tagged object as well as to define certain behavioural patterns of the movement of the objects which can be used to improve the monitoring accuracy of these systems.