DFT - based high resolution frequency estimation using three samples

ABSTRACT: Estimation of the parameters of a complex sinusoid in noise usually consists of two steps; a coarse frequency estimation found by applying the N-point DFT of an N length input. Then a search method is applied around the peak frequency. Different search methods can be applied around the peak frequency. In this thesis, we try to compare the method proposed by Candan with the Jacobsen, Macleod and Quinn’s estimators. The performance measure of these algorithms will be compared in terms of the Cramer-Rao lower bound. Different experiments have been implemented with different number of observations. Simulations show that as the number of observations becomes larger, these methods converge to the Cramer-Rao lower bound expressed as RMS error. However, Candan’s method has shown the best performance among all other algorithms. Keywords:Frequency Estimation, DFT, Jacobsen Estimator, Quinn Estimator, Macleod Estimator, Cramer-Rao Lower Bound. …………………………………………………………………………………………………………………………