Sparsifying time-frequency distributions for gravitational wave data analysis

We examine the potential of sparse time-frequency representations for gravitational wave data analysis, as regards: i) characterizing the fine structure of the nonstationary disturbances (glitches) of instrumental/environmental origin affecting the interferometric detectors, and ii) improving the network detection of unmodeled and nonstationary gravitational wave signals coexisting with glitches. These goals are achieved by elaborating on classical and new results about time-frequency and sparse representations. We propose to use a compressed sensing technique (which enables exploiting time-frequency sparsity), in conjunction with an adaptive smoothing of the Wigner-Ville distribution (which enables exploiting the different coherency properties of the different signals of interest).