Studies of the Unbiased Fir Filter for the Time Error Model in Applications to Gps-based Timekeeping

The unbiased FIR filter is investigated to estimate the time interval error (TIE) K-D polynomial model of a local clock in GPS-based timekeeping in the presence of uniformly distributed sawtooth noise. An estimation algorithm is proposed and applied to the GPS-based TIE measurements of a crystal clock without using the sawtooth correction. Based upon this, we show that the TIE estimates fit actual values with an uncertainty of the GPS time. It is also demonstrated that estimates of the fractional frequency offset fit the measurements with the frequency shifts present in the reference rubidium source and the 1 PPS signal of a GPS receiver used.