An event-driven Kalman filter for improved positioning estimation in RFID tracking systems

Radio Frequency Identification (RFID) is a topic of increasing research interest. Presently, the largest application of RFID technology is for inventory control in supply chain management systems. To improve the efficiency of logistics operations, RFID technology has also been under consideration for real-time position tracking. This thesis develops the novel use of a Kalman filter for this application. Unlike conventional state estimation problems, determining position from RFID information occurs as non-uniformly distributed measurements. That is, measurements are acquired whenever the target object passes a reader terminal with no information available at other times. For this application, an event-driven asynchronous discrete-time state position estimator is developed. General design guidelines are developed and the performance is evaluated through a detailed simulation that accounts for RFID read-rate variations as well as microprocessor scheduler issues for implementation with a real-time operating system.