Multiple-model hybrid particle/FIR filter for indoor localization using wireless sensor networks

摘要

This letter proposes a new state estimator called the multiple-model hybrid particle/finite-impulse-response (FIR) filter (MMHPFF) for indoor localization using wireless sensor networks. In the proposed hybrid filtering algorithm, the multiple-model particle filter has the role of the main filter, and it overcomes uncertain process noise problems arising from the use of the constant velocity (CV) motion model in indoor localization. In addition, the multiple-model FIR filter is used as an assisting filter to overcome particle filter failures owing to the sample impoverishment phenomenon. Indoor localization simulations demonstrated that the proposed MMHPFF is more accurate and reliable than conventional algorithms.