Improving pedestrian navigation performance with robust methods for low-cost multi-GNSS

Low-cost GNSS technology has recently expanded as an alternative to geodetic receivers in many applications, including pedestrian navigation, autonomous vehicles, atmospheric monitoring, and precision agriculture. Considering the capabilities of the receiver in terms of power consumption, size, and cost, together with the ability to record multi-GNSS observations, the low-cost receivers and patch antennas have appealed to many users. However, there exist some challenges to be addressed. Such hardware is inherently sensitive to the multipath effect, and the noise level in the observations is relatively higher, resulting in a lower carrier-to-noise density ratio (C/N0). More outliers are encountered for kinematic applications in challenging environments such as urban areas. Therefore, both realistic stochastic modeling (e.g., C/N0-dependent) and the identification of outlier observations are crucial issues for achieving reliable positioning when low-cost GNSS hardware is used. This study aims to investigate pedestrian navigation performance using low-cost GNSS and to enhance positioning accuracy through the implementation of the improved Single Point Positioning (SPP) algorithm, thanks to the code sequence. The technique employs a powerful quality control scheme to mitigate outlier observations. The method consists of two main steps: (i) If the products or measurements used for epoch-by-epoch solutions are troublesome for certain satellites, the median absolute deviation (MAD) method is applied to eliminate these observations. (ii) The remaining observations are then reweighted using a standardized residual-based Institute of Geodesy and Geophysics (IGG) III method during the least-squares. A kinematic test experiment was conducted to validate the usefulness of the approach for which observations were collected from four satellite systems with a 2-s sampling interval of approximately 20 min using a low-cost GNSS antenna (u-blox ANN-MB-00-00) and receiver (u-blox ZED-F9P). The multi-GNSS SPP solution with code observations of GPS L1, GLONASS G1, Galileo E1, and BDS-3 B1 frequencies was performed in this dataset. A GNSS station very close to the study area was used to obtain the reference trajectory with the post-process kinematic method. Analyzing only the fixed coordinates together with the corresponding SPP solution coordinates, resulted in an RMS value of about 0.50 m achieved in the horizontal component. Results showed how the utilization of proposed techniques can enhance basic SPP solutions that yield meter-level horizontal positioning accuracy. Moreover, the suggested technique improved multi-GNSS SPP solution RMS values by 33% in the horizontal and 19% in the vertical component compared with solutions without outlier detection. A comparison was also made using the findings from two distinct software packages to verify the consistency of the outcomes. The results of the evaluation indicate that the SPP algorithm exhibits comparable performance to that of the other software and validates the effectiveness of the employed technique. Finally, the GPS/GLONASS/Galileo/BDS-3 SPP, exhibiting a 3D RMS value slightly better than 2 m for pedestrian navigation, illustrates the capabilities of low-cost GNSS technology.