Feature extraction of intra-pulse modulated radar signals using time-frequency analysis

Abstract

This thesis applies time-frequency transformations to radar signals. Specifically, it considers the feasibility of applying time-frequency transformations to extract the intra-pulse modulation parameters of radar signals. In this work, we consider radar signals with analog pulse compression; specifically linear or hyperbolic intra-pulse modulation. Several time-frequency transformations are investigated to identify which one gives the most accurate image representation for signals in noisy environments. Next, image processing techniques are applied in conjunction with an adaptive curve fitting method, for the hyperbolic modulation scheme, to extract the parameters of the frequency equation. Results show that for the linear chip case the frequency equation can be estimated with small error down to SNR equal to -10dB. The proposed method for the hyperbolic chirp modulation is less immune to noise degradation and it can be used down to SNR level equal to 2dB.