High-precision Frequency Estimation Algorithm

Frequency estimation is an important part of digital signal processing, to estimate thefrequency of sinusoid drowned in noise is a classical subject of signal processing.Currently, high-precision frequency estimation has been successfully applied to radardetecting, sonar and earthquake monitoring, bridge vibration testing and electroniccommunications technology, therefore, the study of high-precision frequency estimationalgorithm has an important theoretical and applied value.The paper is research on several types of high-precision frequency estimationalgorithms, including the amplitude ratio method, phase difference method, spectrumthinning method and autocorrelations assisted method, compare the algorithms'performance based on the analysis of its principle and features, get the relation ofroot-mean-square deviation and relative frequency deviation, as well as the signal-to-noiseratio by Monte Carlo simulation, and then compare the estimated root mean square error tothe Cramer-Rao lower band. Meanwhile, the paper advances some improvedhigh-precision frequency estimation algorithms. The improved parabolic interpolationmethod is a supplement to Rife algorithm, of which performance is poor when the signalfrequency is near to the discrete frequency. The interpolation method related withautocorrelation pretreatment, can be effectively suppressing noise and improving theperformance in low SNR environment. In addition, the paper reduces the computation offrequency estimation algorithm based on frequency offset correction.The paper analysizes a simple and effective Pulse-Pair algorithm with lesscomputation, the thinking of Pulse-Pair method may have a reference to many spectrumresearch scholars.