Identifying a low-frequency oscillation in Galileo IOV pseudorange rates
- First Online:
- 971 Downloads
Galileo, the European global navigation satellite system, is in its in-orbit validation phase and the four satellites which have been available for some months now have allowed a preliminary analysis of the system performance. Previous studies have showed that Galileo will be able to provide pseudorange measurements more accurate than those provided by GPS. However, a similar improvement was not found for pseudorange rate observations in the velocity domain. This fact stimulated additional analysis of the velocity domain, and, in particular, an unintended oscillatory component was identified as the main error source in the velocity solution. The magnitude of such oscillation is less than 10 cm/s, and its period is in the order of few minutes. A methodology was developed to identify oscillatory components in the Galileo IOV pseudorange rate observables, and it was verified that the measurements from Galileo IOV PFM and Galileo IOV FM2 are affected by a small oscillatory disturbance. This disturbance stems from the architecture adopted for combining the frequency references provided by the two active clocks present in the Galileo satellites. The issue has been solved in Galileo IOV FM3 and Galileo IOV FM4, and the oscillatory component has been eliminated. We also propose a methodology for removing this unwanted component from the final velocity solution and for determining the performance that Galileo will be able to achieve. The analysis shows that Galileo velocity solution will provide a root-mean-square error of about 8 cm/s even in the limited geometry conditions achieved using only four satellites. This shows the potential of Galileo also in the determination of user velocity.
KeywordsGalileo In-orbit validation (IOV) Oscillators Velocity analysis
Galileo, the European global navigation satellite system (GNSS), is in its validation phase, and four in-orbit validation (IOV) satellites allow the analysis of the system’s stand-alone performance. Several research groups have investigated the performance of Galileo positioning both in single point (SP) mode (Gioia et al. 2014) and using a real-time kinematic (RTK) approach (Odijk et al. 2014). The potential of Galileo has been clearly established, and, in particular, it has been shown that Galileo will be able to provide pseudorange measurements more accurate than those from GPS. However, a similar improvement was not found for pseudorange rate observations and consequently in the velocity estimation (Gioia et al. 2014). This result motivated the additional analysis described here, where the IOV pseudorange rate observations and Galileo stand-alone velocity solution are further studied.
The analysis showed that small oscillatory disturbances are present in the velocity solution computed for a static Galileo receiver. Moreover, the study of the different components used for velocity computation indicated that the sources of such oscillations are the Galileo pseudorange rate measurements. For this reason, a methodology was developed to identify oscillatory components in the Galileo IOV pseudorange rate observables, and it was shown that only the measurements from Galileo IOV PFM and Galileo IOV FM2 are affected by this disturbance.
The presence of small oscillatory components in the pseudorange rate measurements from Galileo IOV PFM and Galileo IOV FM2 has been verified using a modified experimental setup involving two Galileo-capable receivers. In this way, it was possible to obtain synchronous measurements from two independent sources, and the phenomenon was observed in the measurements from both receivers. Finally, high rate data from the International GNSS Service (IGS) (Dow et al. 2009) were used, and the analysis was repeated on pseudorange rate measurements from IGS station WTZ3, from Wettzell, Germany. Also in this case, oscillatory components were clearly observed. Moreover, the analysis of datasets collected in different days proved that the centre frequency of the oscillatory component is time varying.
The presence of small oscillatory components in the measurements from Galileo IOV PFM and Galileo IOV FM2 was discussed during the IOV review event organized by the European Space Agency (ESA) in October 2013. From the discussion, it emerged that the disturbances observed are due to the approach adopted for combining the frequency references produced by the on-board clocks of the Galileo IOV PFM and Galileo IOV FM2 satellites. Galileo is the first GNSS which will use several on-board clocks for the generation of time and frequency reference signals (Rochat et al. 2012). In particular, two passive hydrogen masers (PHMs) and two rubidium frequency standards (RFSs) clocks are installed on the IOV satellites (Rochat et al. 2012). The PHMs are used as primary clocks, whereas the RFSs are activated in case of failure of the primary devices. Thus, two frequency references are available at any time, and a combining algorithm is needed to generate a single timing signal. The first version of the combining algorithm generates a spurious oscillation at the frequency of about 6 Hz. When the measurements are taken at a frequency of 1 Hz, the spurious oscillation is aliased back to a very low frequency. The disturbances observed are the result of this aliased spurious oscillation.
The identification of disturbances in the measurements is fundamental for the correct interpretation of the results obtained using the products provided by GNSS satellites. For example, Benton and Mitchell (2012, 2014) showed that phase measurements from IIR GPS satellites are sporadically affected by clock anomalies which could be interpreted as strong ionospheric events. The analysis provided in (Benton and Mitchell 2012, 2014) allows the exclusion of ionospheric phenomena and the correct interpretation of the experimental results obtained using GNSS observables. The analysis presented here allows the correct interpretation of the results obtained using Galileo pseudorange rate measurements.
The oscillatory disturbances are not present in Galileo IOV FM3 and Galileo IOV FM4: Future Galileo satellites will be able to fully exploit the benefits brought by the Galileo atomic clocks. For this reason, Galileo performance should be analyzed in the absence of such oscillatory components. With this in mind, the velocity analysis performed by Gioia et al. (2014) has been extended, and a methodology based on a two pole notch filter (Borio et al. 2008) has been suggested for removing such disturbances from the velocity solution. The analysis performed in (Gioia et al. 2014) has been repeated using filtered velocity components, and the potential of the Galileo system has been proven also with respect to the determination of the user velocity.
Although the E1 signal is considered for the experimental analysis, oscillatory components can be also observed in measurements from other frequencies.
Experimental setup and velocity computation
In this section, the experimental setups used to collect Galileo observables are described along with the algorithm adopted for the velocity computation.
Two different setups were adopted in order to perform different types of experiments. The methodology described in (Gioia et al. 2014) was initially adopted for the pseudorange rate and velocity analysis. A Javad RingAnt-G3T was mounted on the roof top of the European Microwave Signature Laboratory (EMSL) in the Joint Research Centre (JRC) premises in Ispra, Italy. The position of the antenna was carefully selected in order to minimize multipath effects. The antenna was then connected to a Septentrio PolarRxS receiver able to simultaneously collect GPS, GLONASS, and Galileo measurements on several GNSS bands. The position of the antenna was carefully surveyed using double difference carrier phase positioning. This information was used to compute the user velocity as described below. With this calibrated setup, it was possible to collect several days of data which were used for the characterization of Galileo observables as discussed by Gioia et al. (2014).
A second setup involving two Galileo-capable receivers was then used to validate and further investigate the results obtained with the Septentrio PolarRxS receiver. The use of a second receiver allows one to verify that the oscillations observed were not artifacts created by the Septentrio PolarRxS receiver or by other local effects.
These two setups were used to collect Galileo measurements and compute the velocity solution according to the methodology detailed below. In particular, GNSS receivers are able to provide Doppler measurements representing the frequency shift produced by the satellite/receiver relative motion (Kaplan and Hegarty 2006). Using Doppler observables, GNSS receivers are able to compute the three-dimensional user velocity; the algorithm adopted here for velocity estimation is developed in the East, North, Up (ENU) frame.
Pseudorange rate pre-filtering
As discussed in the introduction, the oscillations observed in the velocity solution motivated a thorough analysis of the pseudorange rate measurements which are used for the velocity computation. It was verified that a direct inspection and a frequency domain analysis of the observables do not allow a clear identification of oscillatory components in pseudorange rate measurements. In particular, pseudorange rate variations are mainly due to the satellite motion (Hoffmann-Wellenhof et al. 1992). These variations are several orders of magnitude higher than the oscillations observed in the velocity domain. Over short time intervals, pseudorange rate variations due to satellite motion can be approximated by a polynomial curve and thus can be removed through filtering. For this reason, a pre-filtering stage was designed to remove these components.
T0 is the period corresponding to such oscillation. The analysis also confirms that the filtered pseudorange rates from Galileo IOV FM3 and Galileo IOV FM4 do not show such behavior, indicating that these unwanted components only affect the first batch of IOV satellites.
It is noted that different approaches could have been used to isolate the oscillatory components described above. For example, the polynomial variations induced by the satellite motion could have been predicted from the satellite and user velocities and positions, and these variations could then have been removed from the pseudorange rate measurements. Similarly, fitting of low order polynomials can be used as suggested by de Bakker et al. (2009).
Filtering was preferred, in order to remove any reliance on external data such as satellite ephemerides: With filtering, only the actual measurements are employed and no other data, which could introduce additional errors, are used.
In this section, the second experimental setup detailed in Fig. 1 has been used to further validate the presence of oscillatory components in the Galileo pseudorange rates.
The data used for the evaluation of the PSDs in Fig. 5 indicate that the oscillatory components observed in the measurements from Galileo IOV PFM and Galileo IOV FM2 have a time-varying centre frequency. This fact clearly emerges by comparing the results obtained in Fig. 4. Also in this case, no such component was observed in the data from Galileo IOV FM3 and Galileo IOV FM4.
In addition to the use of a modified experimental setup, data were downloaded from the IGS (Dow et al. 2009) and used for the analysis. In this way, it was possible to use data from a different site, obtained in a different way. Station WTZ3 located in Wettzell, Germany, is equipped with a Javad Delta receiver and provides high data rate (1 Hz) Galileo observables in RINEX 3 format. These data were used to further verify the results described above.
These results confirm the findings obtained in the previous sections: Pseudorange rate measurements from Galileo IOV PFM and Galileo IOV FM2 are affected by oscillatory components which slightly degrade the final velocity solution obtained using Galileo IOV signals.
Filtered velocity and Galileo velocity accuracy
As mentioned in the introduction, the oscillations measured in the pseudorange rates are caused by the initial architecture adopted for combining the frequency references provided by the two active clocks present in the Galileo satellites. The combining algorithm has been improved in Galileo IOV FM3 and Galileo IOV FM4, and the oscillatory component is no longer present. Since the problem has been effectively solved, Galileo will be able to provide clean pseudorange rate measurements and improved velocity solutions.
Statistics of the filtered and unfiltered Galileo velocity errors
It is worth noting that Gioia et al. (2014) have already evaluated the velocity solution using Galileo observables without filtering. Their results were affected by the oscillations defined in this research, but the cause was not identified.
Statistics of the velocity errors: comparison between GPS (with limited DOP) and filtered Galileo solutions
RMS horizontal error (m/s)
Mean horizontal error (m/s)
Maximum horizontal error (m/s)
GPS limited DOP
The values provided in Table 2 confirm that Galileo and GPS have similar performance in the velocity domain. Although the maximum error observed for Galileo is higher than the GPS one, RMS and mean horizontal errors are slightly reduced. As shown in (Gioia et al. 2014), without filtering, Galileo performance seems degraded with respect to GPS.
We proposed a methodology for isolating oscillatory components in pseudorange rate observations and used it for the analysis of Galileo observables. From the analysis, it emerged that measurements from Galileo IOV PFM and Galileo IOV FM2 are affected by small oscillatory components which are due to the approach adopted for combining the reference signals provided by the Galileo on-board clocks. The problem has been solved, and measurements from the second pair of IOV satellites are not affected by such oscillations. Future Galileo satellites will therefore provide enhanced measurements and improved velocity solutions.
A notch filter was adopted to remove the oscillatory components from the Galileo velocity solution and assess the performance that Galileo will be able to achieve. The analysis shows that Galileo has performance similar to that of GPS under similar geometry conditions.
Future work will include the investigation of other Galileo observables and the impact of the oscillations identified on such measurements.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.