Predictive NLOS Detection for UWB Indoor Positioning Systems Based on the CIR

Modern indoor positioning systems are used to track machines or navigate people. Besides localization accuracy, one emerging key requirement is the robustness of a system with regard to changes in the environment. One mechanism to improve this robustness is the detection of Non-line-of-sight (NLOS) situations. In this paper, a novel, predictive NLOS detection concept is introduced. Moving objects induce changes in the Channel Impulse Response. While close-range obstacles are identified by a pulsing first path peak amplitude, objects further away can be detected using a Gabor filter bank. Both features for detection have been investigated in ideal and real-world environments with the results showing clear evidence that our novel method is delivering good-confidence NLOS detection.