7. Computer methods for processing signals and images

V.S. Sobolev, F.A. Zhuravel’
Institute of Automation and Electrometry, Russian Academy of Sciences, Siberian Branch, Russia

Comparison of two algorithms for evaluation of central frequency narrowband normal random process with gaussian spectrum

A comparison of two algorithms of the central frequency estimations of the normal random narrowband process with Gaussian spectrum is performed. The first is based on the method of quasi-optimal evaluations, and the second is realized by averaging the instantaneous frequency estimates. It is shown that quasi-optimal algorithm does not shift the Doppler frequency estimate (i. e. speed estimate), gives a reasonably exact frequency estimates (0.12%) and has high noise immunity, while the second algorithm also does not shift the estimate, but gives much higher (at the order) errors. Its application is only possible with the use of narrowband filtering of the signal itself or its instantaneous frequency.
Full text paper