HOS-based symmetric and asymmetric statistical models of non-Gaussian noise for signal detection optimization*

In the context of digital signal processing addressed to communications, this work focuses attention on the optimization of detection of weak signals in presence of additive independent stationary non-Gaussian noise. In order to optimize the detection performances for low SNR values, the selected binary statistical testing approach consists in a Locally Optimum detector, designed on the basis of new proposed models of non-Gaussian noise probability density function (pdf). The investigated analytical models are expressed in terms of various HOS statistical parameters (of the third and fourth orders). In particular, symmetric (depending on the kurtosis parameter) and asymmetric (in terms of the skewness parameter) pdfs have been studied and mutually compared, in order to describe realistically the non-Gaussian noise and improve detection performances. The various resulting methods have been compared with the Gaussian-hypothesis LO test. Experimental results have shown significant advantages in modelling noise pdf on the basis of HOS parameters; the different methods have been applied for detecting known deterministic test signals corrupted by real ship-traffic-radiated non-Gaussian noise.