Title :
A time scale based on optimum linear prediction
Author_Institution :
Newton Metrol. Ltd., Jerusalem, Israel
Abstract :
A time scale in which adaptive, close to optimum, linear predictors are used to calculate each clock´s weight, is described. This algorithm is currently used to generate TA(INPL) time scale. For each type of oscillator instability there exists an optimal or close to optimum linear predictor. A set of close to optimum predictors is predefined and their prediction is calculated in each algorithm iteration step. The prediction error of the best predictor for each clock is used to assign a weight to the associated clock. The weights are then used to calculate the weighted average time scale. Results with TA(INPL) and post analysis with other laboratories data are presented
Keywords :
atomic clocks; computer aided analysis; frequency stability; iterative methods; measurement standards; prediction theory; TA(INPL) time scale; algorithm; algorithm iteration step; clock´s weight; laboratories data; optimum linear prediction; optimum linear predictor; optimum predictors; oscillator instability; post analysis; prediction error; time scale; Clocks; Frequency; Hardware; Integral equations; Laboratories; Least squares methods; Metrology; Oscillators; Testing; Timing;
Conference_Titel :
Frequency Control Symposium, 1997., Proceedings of the 1997 IEEE International
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-3728-X
DOI :
10.1109/FREQ.1997.638571
