Page 101 - 卫星导航2021年第1-2合期

Shinghal and Bisnath Satell Navig (2021) 2:10 Page 4 of 17

3. Kinematic: Two datasets collected with the phone as carrier-to-noise ratio, data gaps and multipath and
clamped to the car dashboard, driven in a medium their correlation with each other are then analysed in
multipath environment at York University, DOY 85 diferent multipath environments.
and 325, 2019. Collection duration of 41 and 33 min,
respectively with SwiftNav antenna placed on the car Carrier‑to‑noise density ratio and multipath
roof (Fig. 2c). C/N is measured and outputted by the smartphone
0
4. Kinematic: 30-min datasets collected in a high mul- data logger and is dependent on: the power density of
tipath urban environment (DOY 54, 2019) and for- the incoming GNSS signal; reception area and gain of
ested area (DOY 67, 2019) with the phone in hand the receiver antenna; satellite elevation; and the receiv-
while walking and in the pocket while skiing, respec- ing hardware, including antenna, receiver and cables
tively (Fig. 2d). (Braasch and van Dierendonck 1999; Fortunato et al.
2019). Low and irregular C/N values can be attributed
0
Te measurements were processed with the YorkU to the inability of a smartphone monopole GNSS antenna
PPP engine—a complete user PPP processor. Tis to distinguish between incoming right-hand circularly
research focuses on dual-frequency GPS (L1 and L5) polarized signals and refected left-hand circularly polar-
and Galileo (E1 and E5a) PPP processing in the uncom- ized signals. Low signal strength and variations further
bined mode. Te measurements were processed using compound signal multipath. Te following analysis inves-
a Sequential Least-Squares (SLS) flter, as the variabil- tigates these limitations in various realistic environments
ity in the measurement noise for smartphones GNSS and their subsequent adverse efects on positioning solu-
measurements in diferent environments makes pro- tion quality. Figure 3 illustrates C/N as a function of the
0
cess noise tuning in Extended Kalman Filter (EKF) elevation angle plot for the Xiaomi MI 8 and SwiftNav
processing extremely challenging. An elevation angle Piksi in a medium multipath kinematic scenario.
mask of 10° and a C/N mask of 20 dB·Hz were used, Te received C/N for the smartphone is not infuenced
0
0
as below these thresholds, measurements sufer from by the elevation angle, while for the SwiftNav, a typical
high multipath or have several tens of seconds of data decrease in signal strength with decreasing elevation
gaps. Also, choosing an extremely high C/N mask such angle is observed. Te duration of data collection is about
0
as 30 or 35 dB·Hz results in several satellites getting 20 min and hence, there is a lack of data at all elevation
rejected when data are collected in realistic environ- angles. Due to the short time of observation for each sat-
ments, further reducing the available satellite count for ellite and limited number of satellites being tracked for
processing. Table 1 discusses the diferent PPP process- the entire duration of data collection, there are gaps in
ing parameters deployed in the YorkU PPP processing the elevation plot. Te average C/N of the smartphone
0
engine. Several measurement quality parameters such L1 signal is 23% lower than that of the reference receiver.

Table 1 YorkU-PPP engine processing parameters for smartphones

Processing parameters YORK U GNSS PPP engine settings
PPP processing mode Uncombined dual-frequency
Estimator Sequential least squares
Antenna corrections International GNSS Service (IGS) Antenna Exchange Format (ANTEX)
Satellite orbits and clocks CNT-Centre National d’Etudes Spatiales (CNES)
Elevation mask 10°
C/N mask 20 dB·Hz for smartphone, 15 dB·Hz for Piksi
0
GNSS system GPS, Galileo
Observations processed L1, L5, E1, E5a
Measurement data format RINEX 3.03
Ionospheric mitigation Slant ionospheric delay estimation
Using Global Ionospheric Maps (GIM’s) as pseudo-observations in the
uncombined flter to mitigate and estimate the slant ionospheric
error
Tropospheric modelling Hydrostatic delay: Davis Global Pressure/Temperature (GPT)
Wet delay: Estimated
Mapping function: Global Mapping Function (GMF)