RSS and Phase Kalman Filter Fusion for Improved Velocity Estimation in the Presence of Real-World Factors
A.J. Hoffman, Nico-Paul Bester
Abstract
Motion estimation of RFID tags has been shown to be possible using phase information, while position estimation is possible using received signal strength (RSS). This article combines phase and RSS information using a Kalman filter as sensor fusion technique. It compares the combined RSS and phase fusion Kalman filter technique for motion estimation with the phase only Kalman filter, exponential smoothing filter and raw phase motion estimation excluding a Kalman filter. The influence of frequency hopping broadcasting regulations on the performance of these techniques is investigated, an aspect that has not been properly addressed before. The primary factors found in real-world scenarios that have an effect on motion estimation techniques are identified as multipath reflections and the presence of multiple tags. The influence of such non-ideal environments are characterized and specific vulnerabilities of these techniques are investigated. Finally, the shortcoming of these techniques as well as advantages and possible environments where this technique can be successfully used are discussed.