Abstract: | The reference time scale for all scientific and technologic applications on the Earth, the Universal Coordinated Time UTC, must be as stable, reliable and accurate as possible. With this in view the BIPM and before it the BIH, have always calculated and then disseminated UTC with a delay of about 80 days. There are three fundamental reasons for doing this. i) It takes some weeks for data, gathered from some 200 clocks spread world-wide, to be collected and for errors to be eliminated. ii) Changes in clock rates can only be measured with high precision well after the fact. iii) The measurement noise originating in time links, in particular using Loran-C, is smoothed out only when averaging over an extended period. Until mid-1992, the ultimate stability of UTC was reached at averaging times of about 100 days and corresponded to an Allan deviation oy(~) of about 1,5x10^-l4 when compared to the best primary clock in the world, the PTB CS2. For several years now, a predicted UTC has been computed by the USNO (Washington D.C., USA) through an extrapolation of the values [UTC - UTC(USNO)] as published in deferred time by the BIPM. This is made available through the USNO Series 4, through the USNO Automated Data Service, and through GPS signals. Due to the instability of UTC, the poor predictability of the available clocks, and the intentional SA degradation of GPS signals, the real-time access to this extrapolated UTC has represented the true deferred-time UTC only to within several hundreds of nanoseconds. Recently, there have been dramatic improvements in several domains. i) New commercial Hewlett-Packard caesium clocks and active auto-tuned hydrogenmasers, both presenting a remarkable frequency predictability, are now available in timing centres, ii) The widespread use of GPS in common-views, combined with improved performances of contributing clocks, has improved the stability of UTC, namely oy(r) to the order of 8x10^-15or averaging times of 40 days, and this could also help to decrease the delay of access to UTC. iii) The SA is now overcome both for GPS common-views and for real-time access to GPS time. In addition, other time dissemination methods, which are not intentionally degraded, are now very promising. iv) Modern computer capability and improvements in data communication capability allow quicker and more efficient automatic data transmission. By taking advantage of these advances, it may soon be possible to obtain a real-time estimate of UTC, provisionally entitled UTCp, from the 1pps output of a commercial clock maintained at the BIPM. In this project the BIPM clock is steered to a software time scale computed by combining present data with extrapolated frequencies, relative to UTC, from a small ensemble of highly predictable clocks maintained in several timing centres. In this way and with definitive computation of UTC performed every month, it seems feasible to maintain a representation of UTC to within +/- 60 ns (20 ns standard deviation). The physical clock which delivers UTCp will be used to measure timing signals, such as those from GPS, GLONASS, INMARSAT and a hydrogen-maser on 'board the Russian satellite Meteor 3M. Anyone measuring those timing signals could then link local time scales to the estimated UTC, in near real-time, by simple data communication with the BIPM where time corrections between timing signals and UTCp would be continuously available. We are now studying in detail how all this might be done with a view to carrying out some pilot experiments. |
Published in: |
Proceedings of the 25th Annual Precise Time and Time Interval Systems and Applications Meeting November 29 - 2, 1993 Ritz-Carlton Hotel Marina Del Rey, California |
Pages: | 217 - 230 |
Cite this article: | Thomas, Claudine, Allan, David W., "A Real-Time Prediction of UTC," Proceedings of the 25th Annual Precise Time and Time Interval Systems and Applications Meeting, Marina Del Rey, California, November 1993, pp. 217-230. |
Full Paper: |
ION Members/Non-Members: 1 Download Credit
Sign In |