Detection of a random signal in non-Gaussian noise with non-ideal training interference samples

BACKGROUND: In many problems of statistical radio engineering and radiophysics, statistical conclusions are quite often based on both observations and a priori assumptions about the studied case, e.g. in the form of certain distributions in the studied model. The papers generally solve the problem of combining independent channels used to detect a random signal against a random interference of independent intensity under the assumption of normality of all random variables. AIM: To detect a random signal against additive non-Gaussian noise with non-ideal training interference samples based on the maximin decision rule used to test assumptions. MATERIALS AND METHODS: In this paper, a similar issue is studied for non-Gaussian non-stationary random variables with non-ideal training interference samples. RESULTS: The detection problem is solved based on data from 2K independent channels. In this case, n samples of complex amplitudes are made from signal plus noise mixture in primary K channels and sample interference values are derived from auxiliary K channels. The problem is solved using the maximin decision rule to test the H0 assumption against the alternative H1 assumption. CONCLUSIONS: The article reviews an example of detecting a random signal against an additive non-Gaussian non-stationary interference with a probability distribution density described by the Laplace law.