engleski

Signal Sparsity and Compressed Sensing in the Time-Frequency Domain

An idealized time-frequency (TF) representation of a multicomponent signal is sparse by its nature since it is composed of trajectories that describe the instantaneous frequency (IF) law of each component. A byproduct of classical TF localization methods is that TF plane gets corrupted by inner and outer artifacts. However, the signal components are located around the ambiguity function (AF) plane origin, while the unwanted artifacts are located away from it. In order to remove those artifacts, their samples can be discarded or weighted in a proper way, but in doing so the resolution of the TF representation is also affected. In order to avoid the TF resolution loss, sparse signal reconstruction algorithms based on the L1 and the L2 norms have been applied to the compressive sensed (CS) AF, as described in this paper.

Time-frequency representation; Instantaneous frequency estimation; Signal sparsity; Compressive sensing; Basis pursuit.

nije evidentirano