Power law noise identification is the basis for phase noise modeling. It is also necessary for completing frequency stability description. It is the basis for navigation clock offset prediction and time scale algorithm. It is also meaningful for atomic clock control and analysis of impact of phase noise on time synchronization.As the beginning, the description of frequency stability is introduced. Several power law noise identification methods are introduced. They are slop method, B1/R(n) bias ratio method, LAG1 autocorrelation method, estimate by wavelet transform and phase noise separation method. In addition, the advantages and limits of those methods are given.LAG1 autocorrelation method is the main topic of this thesis, which is introduced in detail. LAG1 autocorrelation method couldn't give identification results with satisfaction when the number of sampled data is below 30. In this thesis, a data extend method successfully used in total variance is applied to the sampled data before the LAG1 autocorrelation method is used to power law noise identification. The simulation results for single kind of phase noise and adjacent integer power law noise show that the data extend skill is useful to improve the identification, except for the WPM phase noise.These power law noise identification methods are used to analyze the real time offset serials between Hydrogen Maser and Cesium clock/ Rubidium clock. The results show that the improved LAG1 autocorrelation method gives smoother estimate than the original LAG1, when the sample number is smaller than 30. |