Prediction of atomic clocks behavior for UTC calculation: a statistical analysis based on clock type.

This thesis presents collaborative research conducted at the Bureau International des Poids et Mesures (BIPM), specifically within its Time Department. The BIPM serves as the international authority on metrology, facilitating global uniformity in measurements and standards. In this framework, the Time Department is specialized in analyzing time differences data, sourced from atomic clocks distributed across laboratories worldwide. These data are optimally combined to establish a stable and precise time-scale, the Universal Coordinate Time (UTC) scale, used as a global time reference. The thesis commences with a comprehensive literature review addressing the current methodologies and their evolution. Thereafter, the study focuses on the implementation and refinement of algorithms for predicting phase error terms. Notably, all clocks contributing to UTC exhibit a deviation from the nominal frequency defining the second, which results in a phase error, as the error in frequency integrates over time. Accurately predicting and correcting the deterministic component of the phase error terms for each clock is essential in maintaining the precision and stability of the UTC time scale. While the current algorithm utilizes a quadratic polynomial, the research explored alternative approaches tailored to specific types of clocks. In fact, there are three main kinds of atomic clocks: cesium (commercial clocks), hydrogen masers (very stable but not accurate) and rubidium fountains (extremely stable and accurate, they lose about 3 nanoseconds in one year). In particular, linear models of the phase deviations were proposed for fountains, addressing the absence of drift which is instead accounted for in the quadratic model. Moreover, a novel method for estimating the linear coefficient, employing the least squares technique instead of the conventional last-first technique, was proposed and investigated. Experimental findings and statistical analysis, detailed within the thesis, provide insights into the efficacy of these methodologies, contributing to the optimization of global time standards.