Adaptive short time fractional Fourier transform for time-frequency segmentation

A novel algorithm is proposed to detect closely placed nonlinear frequency modulation signal components using adaptive short-time fractional Fourier transform (ASTFRFT) combined with time-frequency segmentation. A high-resolution time-frequency distribution is defined by ASTFRFT computed using windows of varying lengths and chirp rates based on spectral kurtosis. The individual signal components are separated by applying a successive iteration of time-frequency coefficients detection. The effectiveness and robustness of this method are evaluated via simulations at low signal noise ratios.