D.S. Grigoryan

In many goals of signal spectral analysis more currency have a questions of targets Doppler, distance and angular superresolution, which coming in compact group. Except of, exists the task of signal from noise countermeasure selection, which source operates from radar antenna mane lobe direction. The results of digital spectral analysis examinations show, that from third to tens parcel of Fourier transformation mane lobe superresolution from 30 dB signal to noise ratio is needed. Therefore radar superresolution demands of procedures, increasing the resolution possibilities of spectral analysis methods in typical for radar missions primary signal to noise ratios low than 30 dB. The purpose of article is tree coherent spectral estimation methods explanation, which increasing the superresolution in comparison with classical autoregressive spectral methods in radar work conditions. The superresolution rising achieves out of proceeds better conditionality of data correlation matrixes, defined by primary data processing methods with data accumulation. The considered coherent estimation methods may be used also in spatial signal processing and in sweeping frequency signals processing, which allows to get distance images. Advantage of considered methods consist of primary signal processing by filters, which have gain differences, defining the signals subspace correlation matrix conditionality decreasing. It gives the weight vector disturbances decreasing. This thesis confirmed by experimental data processing, reflected from two aircrafts, which were under tracing by 9C35M1 combat radar. Those coherent methods may be used by.