High-order statistics for sinusoid peak detection
Karim Drouiche, Messaoud Benidir, Abdelkader Mokkadem · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1995
Detection and estimation of harmonic signals embedded in noise is one of the most encountered problems in the signal processing area. Much research has been done for solving such a problem regarding its importance in many applications. Second order statistics have been used extensively by many authors such as Whittle, Bartlett, Hannan, and Priestley. Each of them proposed a test for harmonic signal detection. However, most of these tests have been proposed under the Gaussina assumption. As a matter of fact, when the the noise is non- Gaussian, statistics of higher order could provide much more information. This is where this paper is directed. We particularly focus our attention on the third and fourth order cumulant methods. Statistical tests based on an extension of the existing tests are used and their efficiency analyzed and discussed.