Kalman Smoothing for better RFID Landslide Monitoring

The use of Radio-Frequency Identification (RFID) in Earth Sciences has been growing in the recent years, notably for landslide monitoring using phase-of-arrival localization schemes. In this article, an Extended Kalman Filtering approach is presented to exploit RFID phase data for landslide displacement monitoring. The filtering is based on a stochastic Langevin equation for the state-space model, introducing a heuristic coupling based on the mechanical continuity of the landslide material. This helps correct measurement biases and deal with missing data in the tracking of multiple tags. The Kalman state covariance matrix is a useful indicator of the tags localization quality. It can be exploited to discriminate true displacements from multipath-induced artifacts. Phase unwrapping is performed implicitly through the state model.