Spectral analysis of irregularly spaced timing data: comparison of several methods

In some applications of time metrology, the measurement cannot be regularly spaced in time. This problem occurs for instance in the case of time transfer using GPS satellites in common view, since it is necessary to wait to see the same satellite from the two clocks being compared. This problem may also be observed in millisecond pulsar timing measurements since pulsars cannot be observed at regularly spaced dates. However, it is important to measure the time stability in these cases. We developed several methods of spectral analysis of irregularly spaced data based upon the multi-variance method. The first class of methods involves different types of interpolation of the missing data, as described in Vernotte et al. (1993). The second class is an adaptation, in the multi-variance framework, of the lowest mode estimator introduced by Deeter and Boynton (1982). The details of these methods are described. In order to compare their efficiency, we applied all of these methods over different types of simulated signals discussed the results