08/04/2005, 15:00 — 16:00 — Room P10, Mathematics Building
Francisco Garcia, ISR
Local stationarity in passive detection of transient signals
Local stationarity is a necessary condition to design passive detectors of transient signals robust to shift errors. It allows for a significant reduction of their computational cost and performance loss. This talk relates the second-order statistics of processes with the sampling theorem and local stationarity of a stochastic process. It is shown how local stationarity can be observed either in the time-frequency plane (Wigner distribution) or the -dimensional power spectrum. A simple method to design locally stationary covariance matrices from data is presented. Real data examples illustrate the advantages of such processors in situations of interest.
Jointly organized with ISR.
