Automated Processing of Low-Cost GNSS Receiver Data

TLDR

Raw GNSS observations from smartphones and tablets introduce significant noise and multipath, making data quality dependent on device and environment and challenging automated processing services such as NRCan PPP. The authors adapted geodetic PPP strategies for low-cost devices by replacing elevation‑dependent weighting with carrier‑to‑noise weighting, providing precise ionospheric corrections with quality indicators, avoiding estimation of residual tropospheric zenith delay, using accurate a priori tropospheric models, and employing geometry‑based quality control. With these modifications, static PPP using Huawei Mate 20X smartphone data converges to centimeter‑level accuracies under favorable signal tracking conditions.

Abstract

The availability of raw observations from smartphones and tablets brings new challenges to GNSS data processing. Low-cost GNSS chipsets, combined with omnidirectional antennas, can lead to measurements highly contaminated by noise and multipath. Therefore, data quality depends not only on the device but also on the environment. Such a diversity is complex to handle for automated GNSS data processing services such as the NRCan precise point positioning (PPP) service. Processing strategies developed for geodetic receivers now require adaptations to be suitable for low-cost devices: 1) carrier-to-noise weighting should replace elevation-dependent weighting; 2) precise ionospheric corrections with meaningful quality indicators should be available; 3) the residual tropospheric zenith delay parameter should not be estimated in the PPP filter, which calls for more accurate a priori tropospheric models; and 4) quality control algorithms should rely on geometry-based rather than geometry-free approaches. With such modifications, static PPP solutions using data collected with a Huawei Mate 20X smartphone can converge to cm-level accuracies under favorable signal tracking conditions.