Detection of weak frequency anomalies for atomic clocks with a Kalman filter

An algorithm presented in a previous paper by the authors detects frequency anomalies. The algorithm was shown to be effective for early detection of weak frequency jumps, and also valid for frequency drift jumps that belong to the class of errors that are most difficult to detect. A modification to the original frequency jump detector algorithm is suggested by the addition of a noise-removal filter. The performance is then compared with the original algorithm and improvement is shown. An analysis of real data from a real satellite is also adopted.