Accurate Self-Localization in RFID Tag Information Grids Using FIR Filtering

Grid navigation spaces nested with the radio-frequency identification (RFID) tags are promising for industrial and other needs, because each tag can deliver information about a local two-dimensional or three-dimensional surrounding. The approach, however, requires high accuracy in vehicle self-localization. Otherwise, errors may lead to collisions; possibly even fatal. We propose a new extended finite impulse response (EFIR) filtering algorithm and show that it meets this need. The EFIR filter requires an optimal averaging interval, but does not involve the noise statistics which are often not well known to the engineer. It is more accurate than the extended Kalman filter (EKF) under real operation conditions and its iterative algorithm has the Kalman form. Better performance of the proposed EFIR filter is demonstrated based on extensive simulations in a comparison to EKF, which is widely used in RFID tag grids. We also show that errors in noise covariances may provoke divergence in EKF, whereas the EFIR filter remains stable and is thus more robust.