Detection of Signals Against Noise in Ultra-Wideband Radar Systems with Subband Processing Information

Abstract

A parametric decision rule for detecting signals against noise in ultra-wideband radar systems with subband information processing has been developed. The decisive rule is based on calculating the likelihood ratio and comparing it with the threshold. Estimates of the expectation vector and the covariance matrix are used as estimates. As a criterion for deciding on the presence of a signal against the background of its own noise, the Neumann-Pearson criterion was chosen, which provides the maximum probability of correct detection of the signal, with a given probability of error of the first row. As a training sample, a sample obtained with a priori absence of a signal, obtained from its own noise, is used. Due to the independence of the frequency channels, with subband processing, the estimate of the covariance matrix obtained at the training stage has a diagonal form with an estimate of the noise level variances in the channels. A procedure has been developed for assessing the level of the decision-making threshold, which is carried out at the training stage. To do this, it is necessary to obtain estimates of the likelihood ratio in the a priori absence of a signal and select the quantile of the normal distribution for a given probability of error of the first kind.