Detection of Underwater Magnetic Transient Signals by Higher

In the study of underwater signals and specially the detection of transients in a noisy background, a lot of second order methods have been developped, and a few bispectrum based techniques more recently, regarding gaussian or stationary noises. Because we are dealing with magnetic signals which are mostly diierent to the acoustic signals, we therefore use methods-in the range of Higher Order Statistics-based on the polycorrelations (cu-mulant sequences) and the WHOS (time-frequency HOS). We have applied them on real underwater magnetic data with no possible hypothesis of gaus-sianity or stationarity to determine their detection abilities in our context.