Parametric spectral analysis of multiple nonlinear chirps using evolutionary algorithms

This paper describes a new parametric method of spectral analysis of complex signals which may contain multiple, overlapping, nonlinear and linear chirps. The method parameterizes the time variation of the chirp frequencies and also provides an estimate of the phase and relative amplitudes of each chirp. The method exploits time windowing of the received signal and uses an evolutionary algorithm to optimize the estimated parameters. The new method works well in high levels of noise corresponding to SNRs of-7dB.