In this chapter, the physical elements in the satellite in charge of generating the timing signal have been presented. First, the AFS generates a reference frequency at fi. Second, a frequency synthesizer in the timing subsystem converts this frequency to the reference frequency F0 as observed on ground. Then the satellite time scale (’real clock’) is created by counting AFS cycles (10.23E6 cycles = 1 second) in the navigation signal generation unit. Finally, the ’Timing Signal’ is created by the navigation unit when encoding the navigation codes over the phase provided by the timing subsystem.
In Section 4.2, the AFS have been reviewed by technology type from the early years to the current state of the art and in the light of future trends. Performance metrics have steadily increased over the last 30 years, beginning with the first dedicated RAFS in GPS adapted from
4.6 Conclusions
German ground technology. Nowadays, new PHM in Galileo and RAFS in GPS have already yielded new possibilities for navigation and POD. Also, new technologies have appeared in the optical domain which have dramatically improved the current AFS performance. Nevertheless, these technologies still need to become mature and to meet space requirements before being embarked in a GNSS satellite. All these items point to the fact that a mixture of different AFS are being used in-orbit, but any application should take into account the diversity and particularities of each AFS in each GNSS.
The frequency distribution unit (FDU) is an important element of the timing subsystem as it allows a user a switch between the different AFS on-board and an adjustment of their fre-quency. AFS noise characteristics can be modified in this step. In contrast, as a-priori satellite clock performance information, the AFS specifications are broadly used. Section 4.3 highlights how this hypothesis can lead to erroneous conclusions. The timing subsystem is composed of two elements: the AFS and the frequency distribution unit. This is particularly true for Block-IIR satellites where the timing signals performance shall be taken from the combination of the AFS(rubidium) plus the settings of the Time Keeping System (TKS), which significantly increase the short term noise of the output signal. The frequency distribution unit can also implement the autonomous detection of any clock anomalies, and subsequently increase the re-liability and integrity of the timing subsystem. This approach was implemented in the TKS and is being reviewed for future Galileo and GPS-III satellites. However, in the short future, a single robust AFS seems to be the most convenient and simplest strategy for navigation satellites; as has been demonstrated by the reintroduction of a similar design from Block IIA in GPS Block IIF for such unit.
Group delays particularly affect GLONASS receivers. A GNSS receiver identifies each GLONASS satellite by its unique frequency allocation, while recognizes other GNSS satel-lites by a common frequency allocation per system and a different code allocation per satellite.
This frequency division implies that the group delays are satellite dependent and, therefore, not fully absorbed by the station clock as a common error for all satellites but by the error budget.
Finally, Section 4.4 presents how frequency dependent phases and group delays are introduced in the timing signal broadcast to the user by the additional elements of the satellite navigation chain. These group delays have a constant part plus a daily variation associated to tempera-ture fluctuations during the orbit period. The same principles are applicable to ground receiver chains. Depending on the signal modulation, the asymmetry of the broadcast signal generates an additional delay in the receiver (geometry dependent) which further limits the accuracy of time transfer. This effect is currently not accounted for by any processing.
5 Methodology applied in geodetic time transfer
5.1 Introduction
The estimation of the offset between the system and satellite time dts is performed by POD adjustment software where the true distance between the satellite and receiver is computed together with other parameters. Code measurements have been introduced as an absolute one-way time transfer from the satellite to the receiver. However the raw measurements do not provide the pure distance between the satellite and the receiver. The code measurement is also often called pseudorange as it includes the true range from the satellite to the receiver plus additional delays caused during the generation, propagation, reception and measurement of the signal. Several strategies exist to overcome the additional delays. They can be modelled with high accuracy (such as relativity), cancelled by linear combinations of different measurements in separate frequencies (as ionosphere), estimated (as clocks), lumped with other terms (as group delays) or simply averaged (as multipath). The clock offset represents only one of these additional delays and is traditionally considered by the geodetic community as a by-product of the satellite and station coordinates estimation.
The basic input data of POD estimations are receiver code and phase measurements. GIOVE satellites allow for free tracking of 7 different modulations (E1A, E1B, E5a, E5b, E5, E6A and E6B) on four separated frequencies. These modulations can carry different information in phase or in quadrature leading to a significant number of tracking configuration possibilities in the receiver with different associated hardware delays. In the standard format Rinex 3.00 a total of 18 different types of tracking codes are allocated to Galileo and 14+2 to GPS, this number being even further increased in version 3.01. Still, despite the numerous modulations and frequencies available only two single frequency measurements are used in POD . Additional measurements are normally ignored and all solutions referred to a basic ionosphere-free combination even by the GNSS service provider (e.g. E1B-E5a for Galileo open service).
The combination of two single frequency measurements into a single quantity as a basic input observation in POD carries some consequences in the timing area. Signals are not aligned at the output of the satellite antenna as explained in Section 4.4 nor in the reception chain, resulting in different group delays associated to each signal. The group delay between the signals gets lumped into the estimated ’ionosphere-free clock’ and consequently the real clock offset is no longer estimated. Furthermore, if two different GNSS constellations with different basic pairs of signals are mixed, the ’ionosphere free clock’ at the station becomes GNSS dependent
and an additional inter-system bias (ISB) needs to be computed. Not all applications rely on dual frequency combinations. Some applications require the estimation of the group delays, such as ionosphere estimations for space weather applications or single frequency users, with the majority of GPS users relying only on GPS C/A code whereas the message refers to P1-P2 combination. Such group delay estimations can provide a way to retrieve the difference between the ’ionosphere-free clock’ and the real signal clock.
This chapter reviews the state of the art of the clock estimation. In the current approaches the group delays are lumped into the ionosphere-free clock and considered constant for each day.
This hypothesis is often ignored and conditions the accuracy of the estimated clocks and any conclusion derived from these estimations. As a consequence, special attention will be given to the group delays. First, methods are briefly reviewed together with IGS product combina-tions normally used as benchmark. Second, the ionosphere-free combination is revised and the other parameters included in the estimated ’ionosphere-free clock’ are identified. From these parameters the group delay is identified as the main bias. Third, the estimation of group delays together with ionosphere estimations is also reviewed. Finally, a practical example of group delay and inter-system bias estimation is presented in Section 5.5 with GIOVE satellites using standard and novel methodologies.