Kalman filter analysis for real time applications of clocks and oscillators

An optimum algorithm has been developed for estimating the time, frequency, and frequency aging of clocks and oscillators. The algorithm is sufficiently general to be used with all types of quartz oscillators and atomic clocks, yet the approach is appropriate for use with an embedded controller in automated systems intended for field applications. The algorithm is based on Kalman filtering techniques and can process either time or frequency calibration data from various sources of different qualities. Data are accepted whenever available and are not required at fixed multiples of fixed sampling time. This algorithm can be used to analyze data already available in many systems in order to improve system timing with no additional hardware. Alternatively, it may be possible to maintain system timing performance while reducing the cost of the clocks. These goals are an important part of the modular intelligent frequency, time and time interval subsystem currently under development.>