New IGS Clock Products: A Global Time Transfer Assessment
November 1, 2002 By: Jim Ray, Ken Senior GPS WorldThe International GPS Service (IGS) has a new suite of clock products available and is continuing to improve their usefulness for practical time and frequency transfer applications. In this month's column, Jim Ray and Ken Senior describe these IGS clock products, use internal repeatability analyses to assess their potential accuracy and stability limits, and compare them with the emerging requirements of the timekeeping community. They conclude that calibration of the internal delays in the GPS receiving equipment will probably continue to set the limit for time transfer accuracy, whereas frequency transfers can already achieve stabilities approaching 10-15 over one-day intervals.
 Figure 1 The geographical distribution of stations included in the IGS combined clock products. The larger, colored symbols denote stations equipped with external frequency standards: H-masers (red), cesiums (yellow), rubidiums (blue). The smaller black dots indicate stations using internal crystals. IGS stations co-located at timing labs are shown as stars. |
GPS has revolutionized the general accessibility of accurate global time and frequency transfer compared with prior terrestrial broadcast systems and the physical transport of clocks. Since the ending of Selective Availability (SA) on May 2, 2000, even users of inexpensive single-frequency, C/A-code-only GPS receivers now have access to GPS Time at a level around 100 nanoseconds, limited primarily by ionospheric propagation error. With a dual-frequency (usually codeless tracking) receiver, the performance improves to the order of 10 nanoseconds at a known location; the leading error sources in that case are inaccuracies in the broadcast GPS orbits and satellite clocks and pseudorange multipath. GPS Time itself differs from the international Coordinated Universal Time (UTC) time scale by similar amounts, up to 40 nanoseconds or so, ignoring leap-second differences. Point-to-point timing comparisons can be further improved to the few-nanosecond range, even for single-frequency users over intercontinental distances, by coordinating observing schedules, exchanging data sets, and forming single-differences to remove the common-mode satellite errors.
During the past two decades, the Bureau International des Poids et Mesures (BIPM) has relied primarily on this "common-view" method to form UTC based on comparisons between clocks at approximately 50 timing laboratories. Despite this important application, the common-view technique is not well suited for general time dissemination due to its coordinated and differential nature. Moreover, the emerging new generation of ultra-cold atomic frequency standards, having one-day stabilities of 10215 or better, requires improving time transfer accuracy to the sub-nanosecond range.
Joint Pilot Project. The products of the International GPS Service (IGS) allow users to exploit GPS in an autonomous, non-differential mode (as well as differentially) to deliver user accuracies about 100 times better than the broadcast navigation system. (Users must additionally observe and analyze the pseudorange and carrier phase data at both GPS frequencies.) This means point positioning good to a few centimeters root-mean-square (r.m.s.) at each measurement epoch or to the sub-centimeter level for one-day integrations.
 Figure 2 The original IGS combined clock estimates for the WSRT station (located in Westerbork, The Netherlands, and equipped with an H-maser frequency standard) for GPS week 1154 (February 17-23, 2002) referenced to a daily linear alignment to GPS Time. We have removed an overall linear trend of 9.7237 ns/day. The large day-to-day discontinuities in offset (time) and rate (frequency) illustrate the limitations of GPS Time as a stable reference time scale. |
In principle, the corresponding time transfer errors could potentially be well below one nanosecond (note that a three-centimeter distance approximately equals 0.1 nanosecond light travel time). This recognition was the basis for establishing a joint pilot project between the IGS and the BIPM, starting in early 1998, to develop and demonstrate the operational capabilities for time transfer.
Hardware. In any GPS-based time transfer technique, the derived user clock readings apply to an internal point within the tracking system. Geodetic analyses of GPS pseudorange and phase data determine the effective receiver "clock", which is at the ionosphere-free phase center of the antenna but offset by the electrical delay to the point in the receiver tracking loop where the observables are measured and time-tagged. To accurately relate the internal clock values to external timing standards, which is essential for most practical applications, one must first determine the instrumental delays and biases within the receiving hardware chain. In general, these delays will not be constant under all circumstances, which greatly complicates matters.
Present methods to measure absolute instrumental delays are accurate only to a few nanoseconds. Differential calibration procedures, which compare a test receiver to an accepted standard unit, promise much better performance, possibly into the sub-nanosecond range. In 2001, the BIPM began a campaign to pursue this idea by circulating an absolutely calibrated receiver as a standard to differentially measure the biases of similar receivers deployed at timing labs. For frequency transfers (as distinct from time transfers), the instrumental requirements are less stringent in an absolute sense but demand that the GPS hardware be highly insensitive to, or well isolated from, environmental change. Such conditions are not unusual at timing labs but are rarely satisfied at typical geodetic facilities.
IGS Clock Products
Satellite clock values are among the "core" products of the IGS. Since its founding, the service has distributed combined solutions for satellite clocks together with the combined satellite orbits. In late 2000, the IGS also began distributing combined clock estimates for a subset of the global tracking network (see Figure 1), as well as for the satellites, both tabulated at five-minute intervals. As many as six independent analysis centers contribute clock determinations, at least two of which are required for each IGS combined value. This method assures quality control. Figure 3 The same WSRT clock data shown in Figure 2 after realignment to the new IGS time scale. We have removed an overall linear trend of 9.0294 ns/day. The observed timing performance is now consistent with that expected of a high-quality H-maser frequency standard. |
Required Consistency. The essential requirement for all the IGS clock products is that they be fully consistent at the centimeter level with the accompanying satellite orbits and the terrestrial reference frame (very closely tied to the International Terrestrial Reference Frame 2000 - ITRF2000). This condition ensures that IGS products can be used instead of GPS broadcast information to attain few-centimeter accuracy. An autonomous user can expect such performance using data from a single, isolated GPS receiver (observing dual-frequency pseudoranges and phases) for both position and clock determinations. The position will be expressed within the highly accurate ITRF2000 frame. On the other hand, the clock will be relative to the underlying IGS time scale, which has historically been only coarsely linked to GPS Time by a daily linear alignment to the broadcast satellite clocks. The time scale of the IGS clock products should ideally be accurately traceable to UTC, in the same way that the IGS geodetic reference frame conforms (and significantly contributes) to the ITRF. We have focused our recent efforts on improvements in this respect.
As with all high-accuracy GPS applications, the IGS relies primarily upon dual-frequency carrier phase observations, which are about 100 times more precise than the pseudorange data. For double-differencing analysis, pseudoranges are not normally even used except to aid in data editing. However, to analyze undifferenced data and determine clock estimates it is necessary to add the pseudorange data in order to permit separation of the otherwise indistinguishable clock offset and phase cycle ambiguity parameters. The quality of the clock estimates is maximized by ensuring the longest possible spans of continuous carrier phase data free of cycle slips, typically three to four hours for each receiver-satellite pair, thus minimizing the number of ambiguity parameters.
1 2 3
RSS
Become a Fan
Join our Network
Follow Us
Please send any technical comments or questions to our webmaster.