Abstract
The existing AT1 algorithm produces a time scale with a fractional frequency variation smaller than that of any clock in the ensemble. We are developing a modification to AT1 that includes the additional desirable features : automatic frequency-step detection, the option to run in an optimal post- processing mode and to run with minimal supervision in non-technical environments. These properties are facilitated by the inclusion of a Kalman-filter estimate of the frequency variance of each clock in the scale. Results are reported from both simulated and real clock data to demonstrate automatic frequency- stcp detection. I. Introduction By sampling an ensemble of clocks, an ideal time scale algorithm would generate time and frequency with more reliability, stability, and frequency accuracy than any one of the individual clocks in the ensemble. In this paper we study an approach to

This publication has 3 references indexed in Scilit: