Page 100 - 卫星导航2021年第1-2合期

Shinghal and Bisnath Satell Navig (2021) 2:10 Page 3 of 17

multiply substantially in realistic environments. Secondly, N -based stochastic model and a measurement predic-
0
smartphone measurement logging applications occasion- tion technique are developed and applied to the datasets.
ally incorrectly compute or format measurements. Tere- Te fnal section of the paper compares PPP static and
fore, to advance positioning accuracy and availability, it is kinematic positioning results with SPP, RTK and smart-
crucial to undertake a detailed analysis of measurement phone internal positioning solutions.
quality, including signal strength, multipath, measure-
ment gaps and cycle slips in these environments and to Raw GNSS measurement collection and analysis
suitably condition the measurements. A Xiaomi MI 8 phone with a BCM47755 GNSS chip was
Te novelty of the presented research lies in its focus used for data collection. It tracks code and carrier-phase
on analyzing and addressing problems with smartphone measurements from GPS (L1/L5), Galileo (E1/E5) and
GNSS measurements in realistic environments. Te two QZSS (L1/L5) and single-frequency signals from GLO-
major outcomes of this paper are: NASS (L1) and BDS. Te SwiftNav Piksi which is a low-
cost receiver was used to obtain cm-dm level reference
1. Analyzing the quality of GNSS measurements in dif- solutions, tracking L1/L2 comparable frequency meas-
ferent multipath environments and addressing non- urements for GPS, Galileo, GLONASS and BDS. Te
continuity and large errors in the PPP positioning smartphone chip costs in the 10 US dollar range, while
solution due to high multipath noise and missing the SwiftNav Piksi chip costs a few hundred US dollars.
GNSS measurements. Te use of a carrier-to-noise Data were collected using the Geo + + Receiver
(C/N ) based stochastic model and an extrapolation- Independent Exchange Format (RINEX) Logger
0
based prediction strategy is shown to reduce posi- (Geo++ 2018) and the GNSS Logger (Diggelen and
tioning errors and increase positioning availability. Khider 2018). Figure 2 depicts data collection using the
2. Comparing the positioning accuracy and avail- (A) Xiaomi MI 8 smartphone and (B) SwiftNav Piksi in
ability obtained by dual-frequency PPP processing diferent multipath environments:
with other techniques such as Real-Time Kinematic
(RTK), SPP and the internal positioning solution for 1. Static: Tripod-mounted phone on the rooftop for one
smartphones. hour on Day of Year (DOY) 225, 2019 (Fig. 2a).
2. Static: Phone taped to the hand of a mannequin,
Te paper begins with a brief description of the receiv- placed on a rooftop at York University, on DOY 146,
ers and loggers used and various data collection scenar- 2019, for two hours (Fig. 2b).
ios. C/N , multipath and data gaps are investigated. A C/
0

Fig. 2 Data collection using (a) Xiaomi MI 8 smartphone and (b) SwiftNav Piksi in diferent multipath environments