Abstract
Integer ambiguity resolution is critical for achieving positions of high precision and for shortening the convergence time of precise point positioning (PPP). However, GLONASS adopts the signal processing technology of frequency division multiple access and results in interfrequency code biases (IFCBs), which are currently difficult to correct. This bias makes the methods proposed for GPS ambiguity fixing unsuitable for GLONASS. To realize undifferenced GLONASS ambiguity fixing, we propose an undifferenced ambiguity resolution method for GPS+GLONASS PPP, which considers the IFCBs estimation. The experimental result demonstrates that the success rate of GLONASS ambiguity fixing can reach 75% through the proposed method. Compared with the ambiguity float solutions, the positioning accuracies of ambiguityfixed solutions of GLONASSonly PPP are increased by 12.2%, 20.9%, and 10.3%, and that of the GPS+GLONASS PPP by 13.0%, 35.2%, and 14.1% in the North, East and Up directions, respectively.
Introduction
Precise point positioning (PPP) technique does not require dense arranged reference stations to achieve static positions with accuracy of the mm–cm level and that of dynamic positioning to the cm–dm level. Thus, PPP has been widely used in geodetic survey, GNSS seismology, precise orbit determination of lowearth orbit satellites, and aerial photogrammetry^{1,2,3,4}. In traditional PPP based on an undifferenced model, ambiguity is usually regarded as a real value. Gabor and Nerem^{5} first proposed the algorithm and the concept of fixing ambiguity of a single station. However, PPP ambiguity fixing is not realized because of the policy of selective availability. Given that the accuracy of orbit and clock corrections is promoted continuously, the technique for PPP ambiguity fixing has improved rapidly in recent years^{6,7,8,9,10,11,12}. Ge et al.^{6} indicated that the uncalibrated phase delays for the single difference between satellites should contain similar fractional cycle biases (FCBs). Reference station network is used to estimate the widelane and narrowlane FCBs based on the single difference between satellites. The estimated FCBs are broadcasted to PPP users, who first need to eliminate the FCB of the receiver using the single difference between satellites. The received FCBs based on the single difference between satellites are then used to eliminate the FCBs of the satellites. In this way, the solution based on PPP ambiguity fixing is obtained. Different from Ge’s method, Laurichesse^{7} et al. and Collins^{8} et al. proposed the integer clock model and decoupled clock model with undifferenced ambiguityfixed solution respectively. In the two methods, the clock biases of the satellite and receiver are utilized to absorb the FCB of the satellite and receiver respectively. In this way, the integer nature of ambiguity can be restored and the PPP ambiguity fixing is realized. This method is known as integerrecovery clock (IRC). Geng^{13} analyzed and compared the methods for fixing ambiguity, namely, the method based on FCB and IRC and proven their equivalence.
When attempting PPP ambiguity fixing, Melbourne–Wübbena (MW) combination observations^{14,15,16} are adopted to fix the widelane ambiguity. However, given that the GLONASS system adopts the signal pattern of frequency division multiple access, the interfrequency code bias (IFCBs) and the interfrequency carrier phase bias (IFPB) will affect the MW combination observations and will further influence the widelane ambiguity fixing. Scholars have comprehensively studied the characteristic of interfrequency bias and its influence on GLONASS ambiguity fixing^{17,18,19,20,21,22}. Their results show that the IFPB is mainly related to the receiver type and has a linear relation with frequency. Residual error can be reduced to a few millimeters after calibration and has hardly influence on widelane ambiguity fixing. However, no obvious linear relationship exists between the IFCBs and the frequency; hence, no unified and effective correction model is known^{18,23,24,25,26}. Reussner and Wanninger indicated that IFCBs are difficult to correct. They proposed a method based on widelane carrier phase observations and an ionosphere model to fix widelane ambiguity. However, their method depends on a highaccuracy ionosphere model. To ensure reliable widelane ambiguity fixing, the ionosphere model must have a prediction accuracy higher than 1.4 total electron content units^{27}. Hence, Geng and Bock recommend that for Europe if only IGS GIMs are available, GLONASS PPPAR is best carried out within an area spanning less than 800 km in both longitudinal and latitudinal directions^{28}. Banville et al. selected two GLONASS reference satellites with adjacent frequency numbers for integer ambiguities definition and thus absorb the linear dependency of the narrowlane IFCBs with respect to frequency^{22}. However, this method disregards the nonlinear parts of the narrowlane IFCBs, even though these nonlinear parts are nonnegligible. Therefore, the IFCBs are critical for processing undifferenced GLONASS ambiguity fixing.
In this study, we assume that IFPB can be precisely corrected in advance based on the exiting studies. To address the influence of the GLONASS IFCBs on widelane ambiguity fixing, we propose an undifferenced ambiguity resolution method for GPS+GLONASS PPP. The data of reference network are used to estimate the IFCBs calibrations and the FCB products, which are then delivered to the client to implement GLONASS ambiguity fixing. The rest of this article is organized as follows: first introduces the ambiguity fixing method for PPP, then elaborates the influence of the IFCBs on GLONASS widelane ambiguity fixing and introduces how the IFCBs are processed in GLONASS ambiguity fixing, then the data of EUREF Permanent network (EPN) are used to verify examples, the corresponding conclusion is presented at the last section.
PPP data processing for ambiguity resolution
In dualfrequency PPP, ionospherefree combination of observations is adopted to eliminate the effect of the firstorder ionospheric delay^{1,2}. The pseudorange and carrier phase observation equation of ionospherefree combination are written as follows:
where i indicates the satellite number; k indicates the satellite number; indicate the pseudorange and carrier phase observation of the ionospherefree combination (meter); indicates the satellite–receiver geometric distance affected by errors, such as relativistic effects and earth tide; c indicates the speed of light; indicates the receiver clock error; λ indicates the wavelength of carrier phase ionospherefree combination of observations; indicates the ambiguity of the ionospherefree combination of observations (cycle), and it can be written as follows:
where indicate the frequencies of carrier phase observation L1 and L2 respectively; and indicate the ambiguities of carrier phase observation L1 and L2 respectively. cannot be solved simultaneously when PPP ambiguity fixing is attempted. The ionospherefree combination ambiguity is then generally expressed in the form of widelane ambiguity and narrowlane ambiguity.
where represents the widelane ambiguity and narrowlane ambiguity, respectively. When attempting PPP ambiguity fixing, widelane ambiguity is initially fixed by MW combination observations, and then narrowlane ambiguity is established by the fixed widelane integer ambiguity and the ionospherefree combination ambiguity. Through narrowlane ambiguity fixing, the fixed solution of ambiguity can be obtained for PPP^{5}.
in equation (3) are actually not integers because these are affected by the FCBs of the satellite and receiver^{16}. Thus, the realvalued undifferenced widelane and narrowlane ambiguities can be expressed as follows:
where represent the widelane and narrowlane integer ambiguities respectively; represent the widelane and narrowlane FCBs in receivers; and represent the widelane and narrowlane FCBs in satellite respectively. The premise of the FCBbased ambiguity fixing method for PPP is using the observation data from the server’s reference network to separate FCB products that cause undifferenced ambiguity to become a noninteger. FCB products are then delivered to the client to restore the integer nature of the client’s undifferenced ambiguity as well as realize the search and fixation of undifferenced integer ambiguities at the client^{6}.
Effects of GLONASS narrowlane IFCBs on widelane FCB estimation
Given that MW combination observations are irrelevant to the satellite–receiver geometric distance, with the firstorder ionospheric delay eliminated^{14,15,16}, MW combination observations are generally adopted to estimate widelane FCBs. The realvalued widelane ambiguity can be expressed as follows:
where indicate the pseudorange and carrier phase observations at L1 and L2 respectively; and indicate the receiver and satellite pseudorange hardware delays respectively. The pseudorange and carrier phase multipath delays and the observation noise can be reduced through the averaging among multiple epochs and are negligible.
As shown in equation (5), widelane FCBs that are estimated by MW combination observations including the effects of the pseudorangerelated hardware delays. For GPS, and are relatively stable. Moreover, is the same for server and client stations, whereas is the same for all satellites tracked at the receiver. Therefore, the characteristics of and will not change if they are combined with and respectively. Accordingly, the following equation can be derived:
where , . For the client, when widelane ambiguity is fixed through MW combination observations, the integer nature of widelane ambiguity can still be restored by using the FCB products^{6}. However, for GLONASS, the pseudorange observation contains IFCBs that are currently difficult to correct^{27}. Accordingly, the realvalued widelane ambiguity can be derived as follows:
where indicates the narrowlane IFCBs^{29} and it should be noted that the IFPB is assumed to have been corrected in equation (7). In GLONASS, the estimation of widelane FCB by using MW combination observations will be affected by the narrowlane IFCBs, as shown below.
however, the IFCBs are related to several factors, such as receiver type and receiver firmware version. Hence, when the server and the client use different types of receivers, variations are observed between the narrowlane IFCBs of the server and that of the client. If the narrowlane IFCBs are ignored, then widelane FCB products estimated by the server will be unable to restore the integer nature of the undifferenced widelane ambiguity at the clientend. Widelane ambiguity fixing will then be affected. Therefore, narrowlane IFCBs processing is the key to GLONASS ambiguity fixing. In the next section, the processing method of narrowlane IFCBs in the widelane FCB estimation will be discussed.
Narrowlane IFCBs processing in widelane FCB estimation for GLONASS
In the FCB estimation, widelane FCB is generally estimated first. After widelane ambiguity is fixed, the narrowlane ambiguity, which is composed of widelane ambiguity and ionospherefree combination ambiguity, is used to estimate narrowlane FCB. Ge et al.^{6} proposed that the substitution of widelane integer ambiguity can avoid the effect of widelane FCB estimation error on narrowlane FCB estimation when the narrowlane ambiguity is composed. According to equation (3), the following equation can be obtained:
however, for GLONASS, the magnitude of the IFCBs can reach several meters at some frequencies^{26,27}. This bias will not only affect widelane FCB estimation, but will also lead to integer deviation in widelane ambiguity fixing. However, the frequency ratio of GLONASS L1 and L2 observations is f_{1}:f_{2} = 9:7. This ratio is substituted into equation (9), which derives the following equation:
the widelane ambiguity’s coefficient is 3.5. For GLONASS, if the fixing of widelane ambiguity generates an integercycle deviation, and when integercycle deviations are odd, the narrowlane ambiguity will have a halfcycle deviation, which will affect the narrowlane FCB estimation. However, when integercycle deviations are even, the narrowlane ambiguity will have integer deviation. This will have no influence on the narrowlane FCB estimation. So, when fixing GLONASS ambiguity, the effects of fractionalcycle deviation (FRAIFCB) and oddcycle deviation (ODDIFCB) on MW combination observations caused by the narrowlane IFCBs need to be calibrated, whereas the absolute value of the narrowlane IFCBs need not be corrected.
Processing method of FRAIFCB on MW combination observations
Considering the effects of the GLONASS IFCBs, the realvalued widelane ambiguity of the GLONASS satellite at station can be expressed as follows:
Fixing the realvalued widelane ambiguity to the nearest integer ambiguity derives the following equation:
where means taking the nearest integer. Assuming that s stations are present in the reference network, and j GLONASS satellites can be observed in each station, the following can be obtained:
where is an order matrix with all elements being 1, is an order identity matrix, and indicates the decimal part of the widelane ambiguity of the satellite at k station. In equation (12), the receiver FCB, satellite FCB, and IFCBs are linearly dependent. In equation (13), the normal equation is rankdeficient and needs the addition of datum constraint equations. The present study selects one station as the base station because only singledifferenced satellite FCB and the IFCBs between satellites make sense. This base station conforms to the following assumptions:
(1) The receiver FCB of base station is 0.
(2) The IFCBs of each satellite that can be observed at the base station is 0.
(3) The summation of the IFCBs of all satellites at the nonbase station is 0.
The datum constraint equation can be expressed as follows:
by adjusting the combined equations (13) and (14), the widelane FCBs of receiver and satellite and the IFCBs can be obtained. Notably, given that equation (12) eliminates the integer ambiguity parameter by rounding, only the FRAIFCB can be obtained and the ODDIFCB needs to be further corrected.
Processing method of ODDIFCB on MW combination observations
Supposing that after correcting the IFCBs for the station , the widelane ambiguity no longer contains ODDIFCB. From equation (10), the realvalued narrowlane ambiguity of satellite can be derived as
After correcting the FRAIFCBs for the station p, if the widelane ambiguity still contains ODDIFCB, then the realvalued narrowlane ambiguity is as follows:
The station is made up of a baseline to evaluate the differences between satellites. The following can then be derived:
where . When the trigonometric function operation is used to eliminate the integer ambiguity parameter, the following can be obtained:
represents the interstation singledifference receiver FCB, which has the same effect on different satellites at the same baseline. This means that if the widelane ambiguity is fixed correctly, the sine and cosine values of realvalued narrowlane ambiguities of different satellites, in theory, are equivalent. But as shown in equation (18), when the widelane ambiguity fixed for a satellite has evencycle deviation, compared with the sine and cosine of other satellites, the values are equal but the signs are opposite. On this basis, the ODDIFCB caused by narrowlane IFCBs on the widelane ambiguity can be calibrated by the following steps:
(1) Selecting one station as the reference station and assume that the widelane ambiguity no longer contains ODDIFCB at the reference station.
(2) Selecting another station which is near the reference station to make up a baseline. The trigonometric function operation is used to the interstation singledifference realvalue narrowlane ambiguities. If equation (18) is obtained, it means the widelane ambiguity contains ODDIFCB at the selecting station and the ODDIFCB should be set to 1.
(3) Selecting other stations to make up baselines with the reference station or the stations which have been calibrated. Repeat the step 2) until all the stations having been calibrated.
To get more precise IFCB calibrations, the coordinates of the stations can be fixed and the postprocess method can be used. As the IFCB is mainly related to receiver type and receiver firmware version^{26}, the IFCB for different receiver type and firmware version can be estimated as well as the FCB estimation and then provided to clients. Anyway, the receiver type or firmware version of the client is probably not included in the reference network. In this situation, we can use the follow steps to obtain the IFCB calibrations.
(1) Assume that the receiver widelane FCB is zero, after correcting the satellite widelane FCBs by the widelane FCB products provided by the server, as shown in equation (12), the FRAIFCB can be easily obtained.
(2) In equation (16), if the satellite narrowlane FCB has been corrected, the remains are only integer ambiguity, receiver FCB, and the halfcycle deviation. Then the trigonometric function operation is used to eliminate the integer ambiguity. Similarly to equation (18), the ODDIFCB should be set to 1, if the values of the sine and cosine are equal but signs are opposite.
Notably, the ODDIFCB at the reference station will be absorbed into the narrowlane FCB. But this will not affect the ambiguities fixing at clientend, as the datum of the wide and narrowlane FCB and IFCB products are unified at the server and clientend.
Experimental validation
The GPS+GLONASS data continuously observed by 45 stations (32 servers, 13 clients) of EPN for day of year (DOY) 335–341 in 2013 were adopted. The station distribution is shown in Fig. 1 and the receiver and antenna types and the firmware version are listed in Table 1. In Fig. 1, the red pentagrams represent reference stations and the blue cycles represent the clients. In this study, the precise orbits and clock error products with 30second sampling interval provided by the European Space Agency were adopted; the errors such as earth tide and phase windup were corrected with the precise model provided by the International GNSS Service (IGS); and the antenna phase center was corrected with the absolute antenna phase center model provided by IGS. The GLONASS IFPB are corrected using the method proposed by Wanninger^{20} while the GLONASS IFCB are processed by the method proposed in this study. After the estimation of widelane and narrowlane FCBs is completed by the server, these FCB products and IFCBs corrections are delivered to the client. To improve the FCB estimation precision, the server fixes station coordinates and the satellite’s elevation mask angle is set to 15°. The client adopts the dynamic PPP model and the same error correction model as the server. The integer nature of undifferenced ambiguity is restored by the FCB provided by the server and the corrected IFCBs. Through leastsquare ambiguity decorrelation adjustment (LAMBDA)^{30}, the widelane and narrowlane ambiguities are searched and fixed, followed by the fixing of ionospherefree combination ambiguity. Thus, the ambiguityfixed solution can be obtained for PPP. This study adopted partial ambiguity fixing strategy to fix the ambiguity and adopts the ratio and probability to check if the ambiguity can be fixed^{31,32}. The thresholds of ratio and are 4 and 0.99 respectively. The calculation equation of probability is as follows^{32}:
Where and represent the float ambiguity and its accuracy respectively, and I is the nearest integer value of b.
To analyze the stability of the IFCB estimation, 30 days data are used to estimate the IFCB (one parameter for each station and each satellite) and the standard deviations (STDs) of the IFCB for each station and each satellite are computed. For simplify data analysis, the average STDs of all 24 satellites for each station are obtained. Figure 2 shows the average STDs of each station. As shown in Fig. 2, the average STDs of all the station are superior to 0.08 cycle and that of most station are superior to 0.05 cycle. This means IFCB is stable and can be consider as the same over 30 days. So the average IFCB over 30 days is consider as the final results for each station. Then, the relationship between different stations is analyzed. Considering that IFCB is related to receiver type and firmware, so in this study, for each receiver type and firmware version, one station is selected as the reference, then the correlation coefficients of the other stations to the reference station are computed. Figure 3 show the correlation coefficient for each receiver type and firmware version. As shown in Fig. 3 the correlation coefficients of the Trimble NetR5 is about 0.85, while the others are all high up to 0.9, which means the stations with the same receiver type and firmware are strongly correlated even if the antenna type is different. Thus, when data are processed, IFCBs of the same receiver type and receiver firmware version can be considered as the same.
Figure 4 shows the widelane FCB estimations of GPS and GLONASS satellites in continuous observation duration of DOY 335–1 in 2013, as well as the standard deviation (STD) of FCB estimations of each satellite in seven days. The GPS and GLONASS widelane FCBs are relatively stable. The STDs of widelane FCB estimations of GPS satellites in seven days are all superior to 0.03 cycle, and the average STD is only 0.012 cycle. GLONASS stability is relatively lower. However, the STD of widelane FCB estimations of each satellite in seven days is superior to 0.09 cycle, and the average STD is superior to 0.05 cycle. It is illustrated that GPS and GLONASS widelane FCBs are relatively stable in a week and can be forecasted for several days.
The residual analysis after the ambiguity fixed by FCB products is an effective method to evaluate the FCB estimation accuracy^{6,10}. This study used FCB estimation products to fix ambiguities of the reference network. After correcting the satellite and receiver FCBs, the ambiguity was fixed through rounding and the residual was analyzed. Figure 5 lists the residuals after GPS and GLONASS widelane ambiguities are fixed and the STD values of residuals in seven days. GPS and GLONASS widelane residuals present normal distribution. Around 99.1% of GPS widelane residuals are less than 0.25 cycle, and around 96.9% of GLONASS residuals are less than 0.25 cycle. After correcting widelane FCB, the widelane ambiguity is very close to the integer value. With 0.25 cycle as the threshold, most widelane ambiguities of GPS and GLONASS satellites can be fixed by widelane FCB products. Based on the STD of widelane residuals in seven days, the STDs of GPS and GLONASS widelane residuals in different days are also closer. The effects of widelane ambiguity fixing among different days are all favorable.
Figure 6 shows the time series of GPS and GLONASS narrowlane FCB estimations and the statistics of the STDs of narrowlane FCB estimations in every 10 minutes. As shown in this figure, the narrowlane FCB estimations are affected by residuals such as clock error and atmospheric delay and change over time in one day, with the STD can reaching about 0.3 cycle. The change of GLONASS narrowlane FCBs is larger than that of GPS narrowlane FCBs. Although narrowlane FCBs are changed considerably in seven days, the change is slow. The STD statistics in 10 min indicate that around 99.6% of the GPS STDs are less than 0.01 cycle, and the largest STD is only 0.012 cycle. Around 98.7% of the GLONASS STDs are less than 0.01 cycle, and the largest STD is around 0.017 cycle. In 10 min, narrowlane FCBs are relatively stable and can be considered constant.
The residuals after fixing GPS and GLONASS narrowlane ambiguities of DOY 335 in 2013 are calculated as well as the STD values of narrowlane residuals in one week, as shown in Fig. 7. For GPS, 98.5% of the residuals of narrowlane FCBs are less than 0.10 cycle, and 99.7% are less than 0.15 cycle; for GLONASS, around 93.5% of the residuals of narrowlane FCBs are less than 0.10 cycle, and 98.7% are less than 0.15 cycle. When GPS and GLONASS narrowlane ambiguities are fixed, the residuals are smaller. This confirm that the narrowlane ambiguity fixing is reliable and the proposed method can effectively eliminate the effects of the GLONASS IFCBs to realize GLONASS ambiguity fixing. However, GLONASS residuals are still relatively large compared with GPS residuals. The first reason may be that the precision of GLONASS orbits and clock error products is slightly worse than the precision of GPS. And the second reason maybe that, for GLONASS, the IFPBs cannot be totally eliminated in this study and the IFPB residuals will affect the accuracy of the narrowlane FCBs.
To further evaluate the effectiveness of the ambiguity fixing method in this study, 13 stations were selected to analyze the performance of PPP ambiguityfixed solutions of GPSonly, GLONASSonly, and combined GPS+GLONASS at the client. The truevalue coordinates of the stations were provided by weekly SINEX solutions. Figures 8, 9, 10, 11 show the positioning time series of PPP ambiguity float and fixed solutions of GPSonly, GLONASSonly, and combined GPS+GLONASS at the SWKI station on DOY 335 in 2013. Ambiguities of the GPS and GLONASS can be fixed using FCB products estimated in this study. After fixing ambiguities, the dynamic PPP accuracies of GPSonly, GLONASSonly, and the combined GPS+GLONASS can all be improved.
Figure 12 shows the average fixing rates of GPS and GLONASS ambiguities at 13 stations in seven days. The fixing rate is around 90% for GPS, but only around 75% for GLONASS. Figures 13, 14, 15, 16 show the average root mean square (RMS) of ambiguity float and fixed solutions of GPSonly, GLONASSonly, and the combined GPS+GLONASS PPP at each station on DOY 335–341 in 2013. After ambiguities are fixed, the positioning accuracy of each station is improved. The average RMS GPSonly PPP at three directions (N, E, U) is reduced respectively from 1.19, 1.26, and 2.65 cm to 0.79, 0.79, and 2.12 cm; the accuracy in the three directions increased by 33.9%, 37.6%, and 20.2%. The average RMS of GLONASSonly at three directions (N, E, U) is reduced respectively from 1.09, 1.30, and 2.65 cm to 0.96, 1.03 and 2.37 cm; the accuracy in the three directions increased by 12.2%, 20.9%, and 10.3%. Compared with that for GPS, the ambiguity fixed for GLONASS contributes less to the accuracy improvement. This result is probably because GLONASS has a lower fixing rate than GPS, and the accuracy of GLONASS narrowlane FCB estimation is relatively lower than that of GPS. For the combined GPS+GLONASS PPP, if only GPS ambiguities are fixed, the average RMS at three directions (N, E, U) is reduced respectively from 0.88, 1.14, and 2.41 cm to 0.78, 0.76, and 2.09 cm; the accuracy in the three directions increased by 11.7%, 33.3%, and 13.3%. And if both the GPS and GLONASS ambiguities are fixed, the average RMS is reduced to 0.77, 0.74, and 2.07 cm, increasing by 13.0%, 35.2%, and 14.1%. It can be seen that, for combined GPS+GLONASS PPP, fixing both the GPS and GLONASS ambiguities contributes less to the accuracy improvement than that fixing only GPS ambiguities. This is because if GPS ambiguities can be well fixed, generally, there will be more than seven ambiguities being fixed. The position accuracy can well be improved. In this situation, fixing GLONASS ambiguities will contributes less to the accuracy improvement.
Conclusions
This study proposes undifferenced ambiguity resolution methods for GPS+GLONASS PPP, which consider the IFCBs estimation. This study also establishes the parameter estimation models for the IFCBs at the server. The client can utilize the satellite widelane and narrowlane FCBs delivered by the server as well as IFCBs products to eliminate the effect of the GLONASS IFCBs. The example result demonstrates that IFCB are very stable over 30 days and the same receiver type and firmware are strongly correlated. Using the IFCB corrections, GLONASS undifferenced ambiguity can be fixed and the success rate can reach 75%. Compared with the ambiguity float solutions, the positioning accuracies of ambiguityfixed solutions of GLONASSonly PPP at N, E, and U directions are increased respectively by 12.2%, 20.9%, and 10.3%; the accuracies of ambiguityfixed solutions of the combined GPS+GLONASS PPP are increased by 13.0%, 35.2%, and 14.1% in the three directions. However, the method proposed in this article requires the IFCBs corrections of different types of receivers. The client need to select the IFCB corrections according their receiver type and firmware version and calculate their own IFCB corrections if their receiver type or firmware version is not included in table list of the IFCB corrections. Furthermore, apart from the selected receivers of Leica series, Nov type, and Trimble series in this article, more types of receivers need to be used for example verification^{33}.
Additional Information
How to cite this article: Yi, W. et al. A method of undifferenced ambiguity resolution for GPS+GLONASS precise point positioning. Sci. Rep. 6, 26334; doi: 10.1038/srep26334 (2016).
References
 1
Zumberge, J. F., Heflin, M. B., Jefferson, D. C., Watkins, M. M. & Webb, F. H. Precise point positioning for the efficient and robust analysis of GPS data from large networks. J Geophys Res 102(B3), 5005–5017 (1997).
 2
Kouba, J. & Héroux, P. Precise point positioning using IGS orbit and clock products. GPS solutions 5, 12–28 (2001).
 3
Kouba, J. A guide to using International GPS Service (IGS) products, IGS report, International GPS Service (2003).
 4
Bisnath, S. & Gao, Y. Current State of Precise Point Positioning and Future Prospects and Limitations. In: Sideris, M. G. (Ed.), Observing Our Changing Earth. SpringerVerlag, New York, pp. 615–623(2008).
 5
Gabor, M. J. & Nerem, R. S. GPS carrier phase ambiguity resolution using satellitesatellite single difference. In: Proceedings of ION GNSS 12th International Technical Meeting of the Satellite Division, Nashville, US, 1999, pp. 1569–1578 (1999).
 6
Ge, M., Gendt, G., Rothacher, M., Shi, C. & Liu, J. Resolution of GPS carrierphase ambiguities in Precise Point Positioning (PPP) with daily observations, J Geod, 82 (7), 389–399 (2008).
 7
Laurichesse, D., Mercier, F., Berthias, J., Broca, P. & Cerri, L. . Integer ambiguity resolution on undifferenced GPS phase measurements and its application to PPP and satellite precise orbit determination. Navigation: Journal of the Institute of Navigation, 56 (2), 135–149 (2009).
 8
Collins, P., Bisnath, S., Lahaye, F., Héroux, P. & Undifferenced, G. P. S. ambiguity resolution using the decoupled clock model and ambiguity datum fixing. Navigation, 57 (2), 123–135 (2010).
 9
Geng, J., Teferle, F., Meng, X. & Dodson, A. Towards PPPRTK: Ambiguity resolution in realtime precise point positioning, Advances in Space Research, 47, 1664–1673 (2010).
 10
Geng, J., Shi, C., Ge, M., Alan, H., Lou, Y., Zhao, Q. & Liu, J. Improving the estimation of fractionalcycle biases for ambiguity resolution in precise point positioning. J Geod 86 (8), 579–589 (2012).
 11
Bertiger, W., Desai, S. D., Haines, B., Harvey, N., Moore, A., Owen, S. & Weiss, J. Single receiver phase ambiguity resolution with GPS data. J Geod, 84 (5), 327–337 (2010).
 12
Li, X. & Zhang, X. Improving the estimation of uncalibrated fractional phase offsets for PPP ambiguity resolution. Journal of Navigation, 65 (3), 513–529 (2012).
 13
Geng, J., Meng, X., Dodson, A. & Teferle, F. Integer ambiguity resolution in precise point positioning: Method comparison. J Geod 84 (9), 569–581 (2010).
 14
Melbourne, W. G. The case for ranging in GPSbased geodetic systems. In: Proceedings 1st international symposium on precise point positioning with the plobal positioning system. 15–19 April, Rockville, pp. 373–386 (1985).
 15
Wübbena, G. Software developments for geodetic positioning with GPS using TI4100 code and carrier measurements. In:Proceedings 1st international symposium on precise point positioning with the plobal positioning system. 15–19 April, Rockville, pp. 403–412 (1985).
 16
Blewitt, G. Carrier phase ambiguity resolution for the Global Positioning System applied to geodetic baselines up to 2000 km. Journal of Geophysical Research: Solid Earth (1978–2012), 94(B8), 10187–10203 (1989).
 17
Wanninger, L. & WallstabFreitag, S. Combined processing of GPS,GLONASS, and SBAS code phase and carrier phase measurements. Proc. ION GNSS2007, Institution of navigation, Fort Worth, Texas, September, pp. 866–875 (2007).
 18
Yamanda, H., Takasu, T., Kubo, N. & Yasuda, A. Evaluation and calibration of receiver interchannel biases for RTKGPS/GLONASS. Proc. ION GNSS2010, Institute of Navigation, Portland, Oregon, September, pp. 1580–1587 (2010).
 19
AlShaery, A., Zhang, S. & Rizos, C. An enhanced calibration method of GLONASS interchannel bias for GNSS RTK. GPS Solutions, 17 (2), 165–173 (2012).
 20
Wanninger, L. Carrier phase interfrequency biases of GLONASS receivers. J Geod, 86 (2), 139–148 (2012).
 21
Sleewagen, J., Simsky, A., Wilde, W. D., Boon, F. & Willems, T. Demystifying GLONASS interfrequency carrier phase biases. InsideGNSS, 7 (3), pp. 57–61 (2012).
 22
Banville, S., Collins, P. & Lahaye, F. GLONASS ambiguity resolution of mixed receiver types without external calibration[J]. GPS solutions, 17 (3), p. 275–282 (2013).
 23
Leick, A., Li, J., Beser, Q. & Mader, G. Processing GLONASS carrier phase observations–theory and first experience. Proc. ION GPS1995, Institution of Navigation, Kansas City, MO, September, pp. 1041–1047 (1995).
 24
Kozlov, D., Tkachenko, M. & Tochilin, A. Statistical characterization of hardware biases in GPS+GLONASS receivers. Proc. ION GPS2000, U.S. institution of Navigation, Salt Lake City, Utah, September, pp. 817–826 (2000).
 25
Zinoviev, A. E. Using GLONASS in combined GNSS receivers: current status. Proc. ION GNSS2005, Institution of Navigation, Long Beach, California, September, pp. 1046–1057 (2005).
 26
Shi, C., Yi, W., Song, W., Lou, Y., Yao, Y. & Zhang, R. GLONASS pseudorange interchannel biases and their effects on combined GPS/GLONASS precise point positioning. GPS Solutions, 17(4), 439–451 (2013).
 27
Reussner, Nico & Wanninger, L. GLONASS Interfrequency Biases and Their Effects on RTK and PPP Carrierphase Ambiguity Resolution. Proc. ION GNSS2011, Institution of Navigation, Portland OR, September, pp. 712–716 (2011).
 28
Geng, J. & Bock, Y. GLONASS fractionalcycle bias estimation across inhomogeneous receivers for PPP ambiguity resolution. J Geod 90(4), 379–396 2015.
 29
Simon, Banville, Paul, Collins & Francois, Lahaye . Compatibility of receiver types for GLONASS widelane ambiguity resolution. Presented at the PPP workshop, 12–14 June 2013, Ottawa, Canada.
 30
Teunissen, P. The leastsquares ambiguity decorrelation adjustment: a method for fast GPS integer ambiguity estimation. J Geod 70 (1), 65–82 (1995).
 31
Teunissen, P., Joosten, P. & Tiberius, C. Geometryfree ambiguity success rates in case of partial fixing. In Proceedings of IONNTM, January, San Diego, CA, pp. 25–27 (1999).
 32
Wang, J. & Feng, Y. Reliability of partial ambiguity fixing with multiple GNSS constellations. J Geod 87 (1), 1–14 (2013).
 33
Dong, D. & Bock, Y. Global Positioning System network analysis with phase ambiguity resolution applied to crustal deformation studies in California. Journal of Geophysical Research: Solid Earth (1978–2012), 94(B4), 3949–3966 (1989).
Acknowledgements
The IGS and the EUREF Permanent Network are acknowledged for providing the productions and tracking data. This work was supported by supported by the National Natural Science Funds (Grant No. 41374034, Grant No. 41404010). This work was also supported by the Open Research Fund of State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing (Grant No. 15P01).
Author information
Affiliations
Contributions
W.Y. wrote the main manuscript text; W.S. wrote the main code of the software and got the experiment results; Y.L. provided the foundational software platform and prepared all the figures; C.S. and Y.Y. designed the experiment and checked the final results.
Corresponding author
Correspondence to Weiwei Song.
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Yi, W., Song, W., Lou, Y. et al. A method of undifferenced ambiguity resolution for GPS+GLONASS precise point positioning. Sci Rep 6, 26334 (2016). https://doi.org/10.1038/srep26334
Received:
Accepted:
Published:
Further reading

Improved Method for GLONASS Long Baseline Ambiguity Resolution without InterFrequency Code Bias Calibration
Remote Sensing (2018)

Determination of near realtime GNSS satellite clocks for the FORMOSAT7/COSMIC2 satellite mission
GPS Solutions (2018)

Calibration of GLONASS InterFrequency Code Bias for PPP Ambiguity Resolution with Heterogeneous Rover Receivers
Remote Sensing (2018)

Estimating the orbit error of BeiDou GEO satellites to improve the performance of multiGNSS PPP ambiguity resolution
GPS Solutions (2018)

Improved method to estimate undifferenced satellite fractional cycle biases using network observations to support PPP ambiguity resolution
GPS Solutions (2017)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.