Robust Statistical Detection of GNSS Multipath Using Inter-Frequency C/N0 Differences.

Multipath detection and mitigation are crucial issues for global navigation satellite system (GNSS) high-precision positioning. The multi-frequency carrier power-to-noise density ratio (C/N-0)-based multipath detection technique has achieved good results in real-time static and low-dynamic applications, and shown better practicability because of the low computational load and the requirement for little additional hardware. However, the classic multipath detection method based on inter-frequency C/N-0 differences directly employs the 3 sigma rule to determine the threshold without considering the distribution of detection statistics and their variation characteristics with elevation angle, and ignores the interference of outliers to the reference functions. A robust multipath detection method is proposed in this paper. The reference functions of C/N-0 differences are fitted using least absolute deviation (LAD) to obtain more accurate nominal values. According to the skew characteristics of the detection statistics, a medcouple (MC)-based adjusted boxplot is employed to determine the threshold. The performance of the new detection method is verified in the multipath environments. The experimental results show that compared with the classic method, the new multipath detector has strong robustness and can respond more accurately to large changes in multipath (MP) combination values at most elevation angles. It is sensitive to short-delay multipath and diffraction, and is an important supplement to multipath detection techniques.