A Multipath Detection Method Using C/N0 Observations From Low-Cost Receivers

Abstract

Multipath effects in global navigation satellite system (GNSS) observations significantly reduce positioning accuracy and are challenging to model and mitigate. We propose three statistical tests for multipath detection using the carrier-to-noise power density ratio (C/ $N0$ ). Each test has certain advantages and disadvantages: the first $S1$ based on the C/ $N0$ is sensitive to an abnormal C/ $N0$ but cannot distinguish multipath and non-line-of-sight (NLOS). The second $S2$ based on the differenced C/ $N0$ between frequencies can detect multipath without being affected by NLOS, but at least two frequencies are needed. Additionally, the combination of $S1$ and $S2$ can be used to identify NLOS. The third $S3$ based on the epoch difference of C/ $N0$ is preferred for kinematic mode. The best performance is achieved when all three tests are used in combination. For the static multipath detection experiment in a harsh environment, a function related to the azimuth is established to mitigate the impact of azimuthal dependence of C/ $N0$ on multipath detection for low-cost receivers. The detection rates for a Trimble Alloy receiver with a geodetic antenna and a low-cost u-blox F9P receiver with a microstrip patch antenna were 83.67% and 73.40%, respectively, while their incorrect detection rates were 10.11% and 20.24%, respectively. For single-epoch static real-time kinematic (RTK) positioning with the u-blox F9P, the proportion of horizontal position error within 0.2 and 2 m were improved from 61.5% to 78.8% and 78.3% to 92.8%, respectively, once multipath observations were identified. For the kinematic experiment, 47.05% and 61.79% of the observations were detected as affected by multipath interference, and 70.95% and 85.27% of NLOS observations were detected via the combination of $S1$ and $S2$ for the alloy and u-blox F9P receivers, respectively. However, due to the sensitivity of low-cost devices to multipath effects, the number of available satellites will significantly decrease during kinematic positioning. In such cases, a weighting method should be applied.