Blind Source Separation and Wigner-Ville Transform as Tools for the Extraction of the Gravitational Wave Signal

Coalescing binaries are credited as being relatively abundant sources of gravitational radiation, with a rich content of physical information. Their signals, apart from (important) complications due to higher-order post-Newtonian corrections, spin-orbit and spin-spin couplings, etc., are so-called chirp signals, i.e. a signal modulated both in amplitude and in frequency. The rate at which the frequency changes depends basically on the chirp mass, a particular combination of the masses of the two objets. It is known that the Wigner-Ville transform is an optimal time-frequency distribution in detecting chirping signals whose instantaneous frequency grows linearly in time. We compare the performance of the plain Wigner-Ville transform and of blind source separation-augmented Wigner-Ville transform. We consider a typical chirp of interest for ground-based gravitational wave (GW) detectors and inject it at a SNR=12 into two independent time series of white Gaussian noise. We show that the blind source separation preprocessing acts as a powerful denoising tool, yielding a significant enhancement in the detection capability of the Wigner-Ville transform alone. We report preliminary results, focused on detection performances, which appear to be very promising; the improvement in parameters estimation will be discussed in a forthcoming paper. The possibility to apply our analysis to a network of GW interferometers is briefly discussed. Finally, we stress the fact the our methods are completely independent on the shape of the signal, and thus have broader applications besides chirp gravitational wave signals.