The Research Of High Resolution Frequency Estimation

The writing of this paper is under the background of additive noise, especially the low SNR, to estimate the parameter of frequency. At first, the history of frequency estimation is reviewed in this paper, and some of the main estimation arithmetics, which include FFT method; maximum likelihood method; modified covariance method; multiple signal classification method; ESPRIT method and so on, are summarized in the second chapter. Then, two signals with close frequencies in WGN are presented, the frequencies are estimated by the methods mentioned above. Computer simulations have been made in this paper and the results of these methods have compared. At last, a frequency estimation method of better capability-ESPRIT is selected, which has high resolution ability in the condition of low SNR, greatly reduces the amount of calculation without the decrease of precision, and needs no scan.The method-ESPRIT selected above is tried to realize on hardware, and the core of it SVD is implemented by the using of CORDIC algorithm on FPGA. All the results in this paper are simulated in MATLAB 7.0 and designed by the devices of Xilinx and its development system ISE 7.1. A summarization of the work of this paper is given at the end of this paper.