engleski

Localized Renyi Entropy Based Sparse TFD Reconstruction

Observing a time-frequency distribution (TFD) as a sparse signal representation provides a significant advantage over the classical view. This approach is possible because many natural occurring phenomena, when observed in the time- frequency (TF) domain, are in fact sparse, thus a TFD is often composed of trajectories describing the instantaneous frequency law of each signal component. Main advantages of the sparse based methods in the TF signal processing include reduction of interference between the components, and components concentration enhancement ; both achieved through a sparse reconstruction algorithm. In this paper, we present a sparse TFD reconstruction algorithm which exploits the information about the number of signal components obtained from the localized Renyi entropy of the TFD. The obtained results are compared to state-of-the-art sparse reconstruction algorithms in terms of the resulting sparse TFD concentration measure and the reconstruction algorithm execution time on a synthetic three-component signal.

Signal sparsity, Compressive sensing, Time-frequency distributions, Ambiguity function, l0-norm, Renyi entropy, Signal reconstruction.

nije evidentirano