Performance and Characteristics of Frequency Estimators for Coherent Laser Radar Systems.

摘要

The aim of the studies in this thesis was to evaluate different methods for estimating parameters of signals originating from coherent laser radar systems, with a potentially very low signal-to-noise (SNR) ratio. Frequency estimating algorithms and values describing the confidence of the estimates have been studied. Many frequency estimators operate on data in the spectral domain. There are basically two different ways of producing estimates, by parametric models or from the data directly. Yule-Walker and Burg are the names of two different estimators both using parametric modeling. Other estimators that have been studied are the Lee, PP (pulse-pair), PG and PG2 algorithms. Lee's method uses a matched filter. The PP uses autocorrelation. The PG algorithms use the periodogram. For the studied signal parameters, the Lee estimator has the best performance. Excluding the knowledge of signal spectral width used by the Lee algorithm we find the PG2 algorithm to be the best. The PP algorithm works well for signals of large spectral widths.