الفهرس 
MAIN REQUIREMENTS FOR GENERATING AND MAINTAINING A NATIONAL TIME SCALE (UTC(K))
The Basic Simple Average Time Scale AlgorithmOnly 14 pages are availabe for public view
 |  | from | 119 |  |  |
| | | from | 119 | | |
Abstract
Improving the frequency stability of the Atomic Time Scale (TA(K)), and hence that of the national time scale (UTC(K)), is a main target for all timekeepers. Coordinated Universal Time (UTC) is the international reference time, and K is the abbreviation given by the Bureau International des Poids et Measures (BIPM) to the timekeeping laboratory. TA(NIS) is the atomic Time Scale (TS) of the National Institute of Standards (NIS), and UTC(NIS) is the national TS of Egypt. The stability of TA(K) can be improved by either increasing the number of atomic clocks in the ensemble or by adding clocks of higher stability. In both cases, a large financial budget is needed which may be unaffordable by some developing economics.
In this thesis, firstly we propose a new method for improving the stability of atomic TS by a factor approaching 50 % with zero expenses by replacing the real clocks in the ensemble by Virtual Clocks (VCs). The VCs are obtained by Downscaling (DS) the frequency offset of the real clocks. The DS process improves the frequency stability of the clocks and hence that of the resultant average TS. Since the down-scaled clocks are correlated to the original ones, then they replace the real clocks and not added to the ensemble. The performance of the proposed method is verified by applying it on the daily clocks comparison data of the Observatory of Paris (OP) published on the time department server of the BIPM. The performance of the proposed method is compared with that of the Kalman filter (KF) noise reduction method presented in the literature. It is found that the proposed method is simpler and acting as a wider bandwidth de-noising filter than the KF noise reduction method for achieving the same performance.
Almost all the timekeeping laboratories maintain the 5071A Cesium (Cs) frequency standard for generating their national time scales. The frequency stability of this standard is limited by different types of noise; especially the White Frequency Modulation (WFM) noise which is the dominant noise source. These noise types affect the stability of the resultant average TS (TA(K)). The KF is still applied till now within the TS algorithm for de-noising and prediction for improving the resultant TS frequency stability. But, this method is very complicated and is based on difficult estimation techniques. In 2013, the Empirical Mode Decomposition (EMD) technique was applied for the first time on Cs atomic clock signal de-noising and frequency prediction.
Secondly, the EMD technique was embedded in the TS generation algorithm for studying its effect on the stability of the resultant TS. Results show that the frequency stability of the resultant TS is improved for averaging times up to nearly 40 days due to EMD. These results are verified by comparing the effect of the EMD to that of the KF on the resultant TS frequency stability. Results show that the frequency stability of the resultant average TS is improved due to using EMD or KF for Cs clock signal-de-noising for averaging times up to nearly 40 days. But, EMD is found to be more effective and simpler than the complicated KF.
Thirdly, a new more effective use of EMD in the TS algorithm for enhancing the frequency stability of national TS was introduced. The EMD was used directly for smoothing atomic clocks errors used in the TS ensemble algorithm without detrending. Then, the smoothed clocks errors were used for building an average TS to improve its resultant frequency stability and hence that of national TS. The use of EMD for smoothing atomic clock error was compared to that of the KF as a famous signal processing technique for the same purpose. Obtained results show that the simple EMD achieves an average TS frequency stability improvement factor of 44% as compared to 41% of the complicated KF. Also, this new proposed use of EMD saves the processing effort and time required for clock detrending step before denoising.
Finally, TA(NIS) was built using two Cs atomic clocks of NIS with the master clock. Obtained results from this simulation were introduced and showed that the frequency stability of TA(NIS) is better than that of the bad stability clock and is very close to that of the other high stability clock. Also, the DS, EMD denoising methods proposed in this thesis for enhancing the frequency stability of TA(K) are applied to TA(NIS). Obtained results show that DS, EMD, and KF noise reduction methods enhance the frequency stability of TA(NIS) for averaging times from 1 day to 10 days.
Before (01/05/2017), it was found from the BIPM monthly Circular T that the TS difference values UTC-UTC(NIS) were between 4000 ns and -8000 ns. These large TS difference values exceed the recommendations of the Consultative Committee of Time and Frequency (CCTF) which recommends that the TS differences UTC-UTC(K) should be kept below ±100 ns.
Moreover, the two steering techniques that are currently used for UTC(NIS) steering were introduced. The first one is based on using the GPS time scale for the prediction of the UTC(NIS) slope tomorrow. The second one is based on using UTCr as a reference for predicting UTC(NIS) slope. Frequency steering values are now calculated using these two techniques and are used for UTC(NIS) steering. Obtained results show that UTC(NIS) was synchronized to UTC in the range of ±40 ns over the last year. This proves the effectiveness of the two UTC(NIS) steering techniques over the last year.