Detecting the number of components in a non-stationary signal using the Rényi entropy of its time-frequency distributions

Authors

  • Dragan Korač
  • Nicoletta Saulig
  • Viktor Sučić
  • Damir Seršić
  • Srđan Stanković

Abstract

A time-frequency distribution provides many advantages in the analysis of multicomponent non-stationary signals. The simultaneous signal representation with respect to the time and frequency axis defines the signal amplitude, frequency, bandwidth, and the number of components at each time moment. The Rényi entropy, applied to a time-frequency distribution, is shown to be a valuable indicator of the signal complexity. The aim of this paper is to determine which of the treated time-frequency distributions (TFDs) (namely, the Wigner-Ville distribution, the Choi-Williams distribution, and the spectrogram) has the best properties for estimation of the number of components when there is no prior knowledge of the signal. The optimal Rényi entropy parameter ? is determined for each TFD. Accordingly, the effects of different time durations, bandwidths and amplitudes of the signal components on the Rényi entropy have been analysed. The concept of a class, when the Rényi entropy is applied to TFDs, is also introduced.

Author Biographies

Dragan Korač

Faculty of engineering, University of Rijeka, Rijeka, Croatia

Nicoletta Saulig

Faculty of engineering, University of Rijeka, Rijeka, Croatia

Viktor Sučić

Faculty of engineering, University of Rijeka, Rijeka, Croatia

Damir Seršić

Faculty of Electrical Engineering and Computing, University of Zagreb, Zagreb, Croatia

Srđan Stanković

Faculty of Electrical Engineering, University of Montenegro, Podgorica, Montenegro

Downloads

Published

2012-03-19