Abordagens para análise e extração de características de radares LPI

Abstract

In this thesis, is presented a comparative evaluation of Time-Frequency Analysis approaches, based on filter banks processing, inserted in a processing methodology that aims the parameters extraction of FMCW (Frequency Modulation Continuous Wave) LPI (Low Probability of Interception) radar signals. Three classes of filter banks are used for comparison purposes: the Quadrature Filters Bank (QMFB) or Johnston Filters Bank, the Cosine Modulated Filters Bank (CMFB) with prototype filter design based on Quadratic Programming (QP) and the Cosine Modulated Filter Bank with prototype filter design based on Least Squares (LS). The determination of the best Time-Frequency Analysis approach is accom- plished with the aid of a processing methodology that, besides analysing the signal in time-frequency domain, extracts its more relevants parameters, by means of image processing techniques, and determines the error performed by this extraction. This error is used as a metric for the determination of the best time-frequency approach. The image processing techniques used in this parameters extraction are based on Mathematical Morphology and Hough Transform.