Detection of the LOS/NLOS state change based on the CIR features

As one of the most promising indoor localization technologies, ultra-wide band (UWB) system attracts lots of attentions due to its high accuracy, low complexity and low-power consumption. However, as one of the most important factors to guarantee accurate position estimation, accurate non-line-of-sight (NLOS) detection remains as a very challenging problem. During the experiments, if the mobile station (MS) move from line-of-sight (LOS) state to NLOS state or from NLOS state to LOS state, dramatic change of the features extracted from channel impulse response (CIR) can be observed. However, if the MS stays in the same state, no big changes can be observed. Based on these observation, four approaches (moving average based, slope based-, wavelet based-, and time series analysis based approach) are proposed to identify the CIR state change (LOS to LOS, LOS to NLOS, NLOS to NLOS and NLOS to LOS). If the initial state is known, the current state can be determined with the accurate state change identification.