Detection of Signals in Non-Gaussian Correlated Noise Derived from Cauchy Processes

摘要

A new processor for detecting a radar target in correlated, non-Gaussian noise is obtained. When this processor and a matched filter are excited with this noise, performance is improved over that of the matched filter alone. The processor is obtained by developing an approximate, bivariate, probability density for the noise and constructing a Neyman Pearson test and then using an approximation to the likelihood ratio obtained from the Neyman Pearson test. The bivariate density was constructed from an underlying Cauchy process and matches the true bivariate density only in the marginals and first two moments.