| Location-based services provided by smartphones have become an important part of informative society in the age of mobile Internet.The smartphone-based high-precision positioning technology has great potential for application in both the mass market and the professional fields.The investigation of Global Navigation Satellite System(GNSS)Precise Point Positioning(PPP)technology on smartphones not only expands the application fields of PPP technology but also strongly promotes the development of positioning technology on smartphones.In 2016,Google released the original GNSS measurement information of Android smartphones,creating conditions for smartphones to achieve highprecision positioning.However,due to the low-cost hardware conditions of smartphones and the complex observation environment,achieving highprecision positioning results remains a significant challenge.On this basis,this work systematically analyzes the GNSS data quality of Android smartphones in different environments and conducts research on the GNSS PPP model and methodology on smartphones.The main research results are summarized as follows.(1)For GNSS signal reception capability,observation data quality and clock stability of Xiaomi MI8 smartphone in four different environments are systematically analyzed.The experimental results show that:(1)The L5/E5 a signals are more susceptible to environmental changes than L1/E1 signals in terms of data integrity rate,and the L5/E5 a signals are more prone to signal interruptions.(2)The Carrier-to-noise Ratio(C/N0)of the L1/G1/B1/E1 signals is significantly correlated with the elevation,while the C/N0 of the L5/E5 a signals is not obviously correlated with the elevation.The Root Mean Square(RMS)value of multipath error for the smartphone is 5 times bigger than that of the Trimble geodetic receiver,and the L5/E5 a signals suffer less multipath effect than the L1/E1 signals.The pseudorange observation noise RMS value of GLONASS is more than 4m,which is 4-6 times of Galileo and BDS,and the carrier phase of GLONASS occurs cycle slips more frequently.(3)The stability of smartphone clock offset is relatively poor due to the standard deviation of its single difference between epochs is 10 times bigger than that of the Trimble receiver.Meanwhile,the inter-system bias is more suitable to be estimated as white noise.(4)Comparing a Xiaomi MI8 smartphone with a Huawei Mate 40 Pro smartphone,we found that the Xiaomi MI8 smartphone is worse in data reception,but better in observation data quality as well as clock stability.(2)An ionospheric-constrained mixed single-and dual-frequency quad-constellation precise point positioning(MSDQ-PPP)model is proposed due to the single-frequency GNSS observations on dualfrequency smartphones still predominate.The MSDQ-PPP is based on the original uncombined observations,with a priori ionospheric information utilized to simultaneously constraint the observations on the L1/G1/B1/E1 and L5/E5 a frequencies to reduce the effect of ionospheric errors,and the GPS/GLONASS/BDS/Galileo constellations are combined to increase the observation redundancy.The experimental results show that:(1)When processing mixed single-and dual-frequency data,the MSDQ-PPP model improves positioning accuracy by 36% and 34% in the horizontal and vertical directions,respectively,when compared to the PPP method with ionospheric-free combination for the dual-frequency data.(2)In an open environment,the positioning RMS error of the Xiaomi MI8 smartphone in static mode utilizing MSDQ-PPP can reach 0.4 m and 0.5 m in the horizontal and vertical directions,respectively,while it drops to 1.5 m and1.7 m for the Huawei Mate 40 Pro smartphone.The positioning RMS error of the Xiaomi MI8 smartphone in dynamic mode utilizing MSDQ-PPP can reach 0.7 m and 1.5 m in the horizontal and vertical directions,respectively,while it drops to 1.8 m and 2.4 m for the Huawei Mate 40 Pro smartphone.(3)In complex environments,the three-dimensional positioning results of the Xiaomi MI8 smartphone are improved by more than 40% compared to both single-frequency-only PPP or dual-frequency-only PPP in static mode,and improved by more than 30% in dynamic mode.(3)To solve the problem that the severe fluctuation of smartphone C/N0 would easily lead to the instability of the C/N0-based weighting model,an observation stochastic model considering the time-varying characteristics of C/N0 is proposed.Combining the advantages of the elevation-based weighting model and the C/N0-based weighting model,a unified weighting model combining elevation and C/N0 is constructed.Experimental results show that the combined stochastic model can effectively improve the PPP precision on smartphones.In an open environment,the combined model can improve by about 23% and about24% in three-dimensional position precision compared with the elevationonly and the C/N0-only weighting model.In a complex environment,the positioning results of Xiaomi MI8 can improve by about 52% and about26%,respectively,and the positioning results of Huawei Mate 40 Pro can improve about 50% and about 19%,respectively. |
