Assessment of GNSS observations and positioning performance from non-flagship Android smartphones

Abstract

Abstract Android smartphone has gained attention in precise positioning applications since it can collect raw observable GNSS (Global Navigation Satellite System) data. Some studies have reported that the positioning accuracy may reach the sub-decimeter level. However, these studies mostly rely on a flagship Android smartphone that is made with better internal hardware, while the use of a non-flagship Android smartphone is not reported for this field. In this study, therefore, we explore non-flagship Android smartphones for positioning applications. We assessed the observable data quality and positioning performance of two non-flagship Android GNSS smartphones of a Samsung M21 and a Redmi Note 7. The data quality assessment includes satellite tracking and carrier-to-noise density ratio analysis. Also, the positioning performance was assessed for Single Point Positioning (SPP) and relative positioning methods in static and open-sky conditions. In addition, the residual properties of GNSS measurements were also evaluated. The results were further compared to the high-grade GNSS device. We found that the observable pseudorange and carrier phase measurements from Android smartphones were about 70 % and 36 % of what high-grade GNSS obtained. Furthermore, within a span of 1 h of observations, a considerable amount of cycle slips, amounting to as many as 518 instances, were noted in the observations from Android GNSS devices. While for the carrier-to-noise density ratio in Android smartphones, it was estimated to be about 15 dB-Hz lower than in high-grade GNSS devices. The spread of the residuals for pseudorange and carrier phase from Android smartphones was estimated to be about ±15 and ±6 m, respectively. The 3D positioning error for SPP was estimated to be about 4.7 m, with a position spread reaching tens of meters. At the same time, the 3D positioning error was calculated to be 4.6 m with the estimated standard error at the centimeter level when using the relative positioning method. To improve the positioning performance, applying a C/N0 mask to the observations become the best solution. The 3D positioning error for the relative positioning method reduces to 2.7 m when applying a C/N0 mask of 30 dB-Hz. The observable data quality of non-flagship Android GNSS devices possibly causes relatively poor performance of positioning applications.