For many years, the time community has been using the precise point positioning (PPP) technique which uses GPS phase and code observations to compute time and frequency links. However, progress in atomic clocks implies that the performance of PPP frequency comparisons is a limiting factor in comparing the best frequency standards. We show that a PPP technique where the integer nature of phase ambiguities is preserved consitutes significant improvement of the classical use of floating ambiguities. We demonstrate that this integer-PPP technique allows frequency comparisons with 1 × 10 −16 accuracy in a few days and can be readily operated with existing products.

In Global Navigation Satellite Systems, the on-board time has to be modelized and predicted in order to broadcast the time parameters to final users. As a consequence, the time prediction performance of the on-board clocks has to be characterized. In order to estimate the time uncertainty of the on-board oscillator, a linear or parabolic fit is performed over the sequence of observed time difference and extrapolated during the prediction period. In 1998, CNES proposed specifications of orbit determination and time synchronization for GNSS-2. The needs of synchronization were specified as the maximum error of the time difference prediction from the extrapolated fit. The purpose of this paper is the estimation of this error for different types of noise. This is achieved by the theoretical calculation of the variances of the drifts coefficients, of the residuals and of the extrapolation errors in the case of quadratic and linear drift models affected by different types of red noises. After the description of the method and its results, examples are given using real data and the predicted extrapolation uncertainties are compared to the real extrapolation errors.

We propose an approach to all-optical frequency standard design, based on a counterintuitive combination of the coherent population trapping effect and signal discrimination at the maximum of absorption for the probe radiation. The short-term stability of such a standard can achieve the level of 10 −14 / ͱ . The physics beyond this approach is a dark resonance splitting caused by the interaction of the nuclear magnetic moment with the external magnetic field.

No abstract

GPS is widely used for time and frequency transfer. To estimate the clock difference between two ground stations, a single baseline solution can be performed using external products for the modelling of the geometrical effects (constellation ephemeris and station coordinates). The baseline solution relies on a single-difference formulation, using code and phase ionosphere-free measurements. The phase ambiguities are usually adjusted as floating parameters. Such solutions give very good results when modelling hypotheses are consistent between the external solution (e.g., GPS orbits) and the baseline solution. However, the frequency bias in the computed clock is very sensitive to discrepancies in the models, and is only observed thanks to the code measurement, with limitations due to the noise. Here, we propose to solve the integer ambiguities on single-difference phase measurements. The advantage is the complete elimination of the clock drifts observed in floating ambiguities solutions. This formulation allows also a reliable continuous connection between overlapping clock solutions (jumps between such solutions can be completely eliminated). Several time transfer results are analyzed and compared to TWSTFT. The methodology has been extended to a network of stations using integer ambiguities on zero-difference measurements. The corresponding results are given for a few European stations.