Improved Multi-scale Chirplet Path Pursuit and its Application in Instantaneous Frequency Estimation of Nonlinear Frequency Modulation Signal

To solve the precise problem of instantaneous frequency estimation for radar nonlinear frequency modulated signal(NLFM) in low SNR, a curve fitting method based on multi-scale chirplet path pursuit(MCPP) is proposed. Firstly, time-frequency plane is divided into binary regions of dynamic time supporting and the multi-scale chirp basis function library is formed. Then, the chirp atom which has the largest projection coefficient is chosen in every regions of dynamic time supporting and all chirp atoms are connected according to the rule of best path. Finally, initial frequency and final frequency of chirp atoms are extracted in order of regions of dynamic time supporting and the least squares curve fitting is adopted for instantaneous frequency estimation. Simulation indicates that the method can improve estimative precision of instantaneous frequency in low SNR.