Generalized matched detection of non-Gaussian background in the frequency domain
Zhenghong Liu · Shengxue jishu · 2007
Detection of a known broadband signal in non-Gaussian noise of unknown power spectral uses generalized matching techniques. In order to reduce variance of power spectrum estimate using the perio-dogram,we use the smoothed periodogram to obtain the estimated spectral of the non-Gaussian to replace the true spectral,and process the signal to do generalized matched filtering under the loss of a constant false alarm rate. Experiments show that performance of the generalized matched detection that estimates power spectrum with smoothed periodogram is only slightly inferior to the true power spectrum,and much better than the result obtained only using a match filter.