Time Scale Estimation by Tracking Parameter Variation
John R. Belcher, Granville Tunnicliffe Wilson · Journal of Time Series Analysis · 2000
A quasi‐periodic time series is sampled at a varying but unknown rate. An autoregressive moving‐average model is fitted to the resulting discrete series and the time variation of its parameters is estimated. The functional dependence of the parameters on the sampling rate is then used to estimate this rate and to reconstruct the true time scale.