Signal detection with higher-order statistics
Chenyang Yang, QU Jian-ming, LI Shao-hong, Mao Shiyi · 2002
A method for detecting deterministic or non-Gaussian stochastic signals in additive noise by applying higher-order statistics (HOS) is presented. This method is not sensitive to the color and the distribution of the noise. The only requirement is that the probability density function (pdf) of the noise is symmetrical. If the waveform of the signal or the spectrum of the colored noise is unknown, the detection performance using a matched filter will be deteriorated, while the HOS method is relatively good.