Characteristics of a real-time ensemble time scale corrected by the KRISS-PSFS using a reduced Kalman filter (Kred) algorithm

Abstract

An ensemble time, which is indicated by TA (temps atomique, in French), was generated every minute by combining four active hydrogen masers with the reduced Kalman filter (Kred) algorithm. The TA frequency was averaged over 12 h and then the mean frequency was steered to the primary and secondary frequency standards developed by KRISS with a micro-phase stepper (MPS). The steered and physically realized TA frequency is called KRISS ensemble time (KET). We investigated the frequency-state estimate outputted from the Kred algorithm and the control signal applied to the MPS, according to the abnormal temperature change of the room where the hydrogen masers were placed. When the frequency of the TA was corrected by the KRISS-F1, the Allan deviation of KET was 1.72 × 10−15 at the average time τ = 1 d, and the variation of [UTC–KET] was stable within 2 ns for 75 d.