Detection of determinate signals with a fixed false alarm probability in the case of autoregression model of interference with unknown parameters

The paper is devoted to investigation of nonadaptive detection of determinate signals with fixed false alarm probability in the case of unknown covariance matrix of interference components. The interference model represents the Gaussian model of random process of the first-order autoregression (AR-interference). Application of the method of projection in the N-dimensional Hilbert space permitted to formulate the conditions, when the low-rank detectors guaranteed fixed false alarm probability at an unknown covariance matrix of AR-interference. The efficiency of this detector is assessed by the simulation method. It is shown that for the first-order AR-interference an acceptable loss (as compared with the Neumann-Pearson optimal detector) can be achieved.