Full-phase time-shift phase difference frequency estimation method and device

Abstract

The invention belongs to the field of digit signal processing, and provides an improved frequency estimation method and device for all-phase time-shift phase difference, capable of improving precision and robustness and lowering signal-noise ratio threshold. The frequency estimation method includes the steps of inputting M=2N-1+L samples, namely x(-N+1-L) to x(N-1), obtaining a first path phase value phi, conducting all-phase FFT for x(-N+1-L) to x(N-1), directly taking the phase of index value K=K2 at peak spectrum as a second path phase value phi2, multiplying the difference of two paths of phase value (phi-phi2) with 1 / delta omega L to obtain the initial frequency bias estimated value epsilon, conducting frequency drift compensation for the initial frequency bias estimated value epsilon to obtain the modified frequency bias estimated value ^ epsilon, and outputting the normalized frequency estimated value as described in the description. The method and apparatus are applicable to digital signal processing.

Description

technical field [0001] The invention belongs to the field of digital signal processing and relates to the problem of improving signal frequency estimation accuracy given finite samples. Specifically, it relates to an improved all-phase time-shift phase-difference frequency estimation method and its device. Background technique [0002] Frequency estimation is one of the most fundamental problems in signal processing, and it is also a key issue in many fields such as vibration analysis [1], array signal processing [2], electromagnetics [3], etc. As we all know, the maximum likelihood estimation of this problem falls on the spectral peak position of the discrete time Fourier transform (DTFT) of the signal, but the ideal DTFT engineering cannot be realized. Therefore, in practical applications, DFT (Discrete Fourier Transform, whose fast algorithm is FFT) is usually used to replace DTFT. In order to overcome the fence effect of FFT, it is necessary to further interpolate [4,5...

AI Technical Summary

Problems solved by technology

[0005] For the original all-phase time-shift phase difference frequency estimator, literature [11] found that its accuracy is related to the degree of frequency deviation of the signal (that is, the closer the frequency deviation value δ is to 0.5, the lower the accuracy), but this literature does not combine apFFT In-depth analysis of the experimental results, and no effective improvement measures were designed

In addition, the most convincing reference to measure the accuracy of the frequency estimator is the Cramer-Rao Lower Bound (CRB) [20], and the literature [9,11] is only for the full-phase time-shift phase difference method in noise The error was tested, and the frequency estimation variance was not compared with the CRB

Images