Stochastic Modeling of Smartphones GNSS Observations Using LS-VCE and Application to Samsung S20

Area of Science:

• Geomatics Engineering
• Satellite Navigation Systems
• Mobile Device Technology

Background:

• Smartphones increasingly incorporate Global Navigation Satellite System (GNSS) receivers, enabling personal positioning and navigation.
• The GnssLogger app allows Android users to record raw GNSS measurements, spurring research into smartphone positioning accuracy.
• Precise Point Positioning (PPP) offers high-accuracy positioning but requires robust functional and stochastic models, especially for mobile applications.

Purpose of the Study:

• To develop more reliable stochastic models for smartphone GNSS observations.
• To investigate the correlation and quality of GPS and GLONASS measurements from smartphones.
• To assess the impact of improved stochastic models on single-frequency PPP (SF-PPP) accuracy.

Main Methods:

• Applied Least-Square Variance Component Estimation (LS-VCE) to double-difference (DD) pseudorange and carrier phase data from Samsung S20 smartphones.
• Analyzed GPS and GLONASS observations on the L1 frequency.
• Evaluated the performance of the developed stochastic models using kinematic SF-PPP data.

Main Results:

• No significant correlation was found between pseudorange and carrier phase observations for GPS and GLONASS on the L1 frequency.
• The quality of GLONASS carrier phase observations was found to be comparable to that of GPS.
• Employing the optimized stochastic model resulted in a 25.1% improvement in horizontal positioning Root Mean Square (RMS) error and a 32.7% improvement in the 50th percentile error for SF-PPP.

Conclusions:

• The developed stochastic models enhance the reliability of smartphone GNSS data processing.
• The findings support the use of smartphone GNSS for precise positioning applications.
• Improved stochastic modeling offers a viable pathway to increase the accuracy of SF-PPP using mobile devices.