Narrowband weak signal detection by higher order spectrum

A study of narrowband weak signal detection by higher order spectrum (HOS) using real signals and real noise, rather than just Gaussian noise, is presented. Noisy real signals are processed using various HOS techniques. We propose to look at the diagonal slices of the bispectrum and the trispectrum as possible substitutes for the power spectrum. The result of applying these slices to weak, real signals is surprisingly good. The dramatic improvements are presented for visual inspection. The performance of the various techniques are then compared quantitatively for different signal-to-noise ratios. The diagonal slices prove to be fast and robust techniques for weak signal detection.