Optimum precise-clock prediction and its applications

We know that optimum precise-clock prediction is equivalent to optimum precise-clock noise prediction. So the optimum of a predicting algorithm depends on the noise model used. There are already several predicting algorithms in history, including polynomial predicting algorithm based on polynomial model, ARIMA predicting algorithm based on ARIMA model, Kalman predicting algorithm based on measurement & state model, etc. But all of these algorithms are not optimum, for the noise models used are not compliant with the power-law model of precise-clock noise. According to the power-law model there are five independent noise parts in precise-clock noise, so the optimum precise-clock prediction should be the sum of the optimum predictions of all five noise parts. This is the principle of optimum precise-clock prediction. In this paper the author presented a useful predicting algorithm, including dynamical separation (according to noise part) of precise-clock noise and the optimum prediction of each noise part, etc. The result of emulation computation on simulated noise series demonstrated that this algorithm is optimum. The algorithm also can be applied in the estimation of present UTC(BIPM) in local time center. The result of emulation computation on three years of UTC(BIPM)-TA(CSAO) series also demonstrated that the algorithm is effective