FREQUENCY DETERMINATION OF PULSE SIGNAL WITH CONSTANT BEAT BY DINT OF RECURRENT USAGE OF FOURIER TRANSFORM

Abstract

The paper proposes an algorithm for frequency estimation of a pulse signal with constant beat under significant distortion conditions with the use of spectral analysis methods. The sequential processing of the signal based on the recurrent usage of discrete Fourier transform is shown. The initial pulse signal is represented as a fragment of the Dirac comb. Its Fourier spectrum also contains a fragment of the Dirac comb. To reduce the effect of leakage due to the application of Fourier transform to the finite sequence, a convolution of the signal with a weight window is used. To increase the signal-to-noise ratio and reduce the side lobe level, the spectrum of the initial signal is taken as the original for Fourier transform with an increased number of spectral lines. The necessity of recurrent transformation usage is brought about by the noise of the initial signal and the resulting spectrum after the first application of Fourier transform. In comparison with a single application of the transformation, this approach provides an opportunity to recognize a useful signal both against the background of Gaussian and aperiodic impulse noise, and against the background of signals containing such noise in the Fourier spectrum. We give a method for checking the local maxima of the amplitudes of the obtained discrete Fourier transform for periodicity. The frequency of a pulse signal with constant beat is determined by the element index with the largest number of fulfillment of the periodicity condition.

Keywords

• Fourier transform recurrent usage
• delta function
• Fourier transform
• Dirac comb
• spectral analysis
• noisy signal
• leakage effect
• Fourier spectrum
• window function
• digital signal processing