Parameter Estimation of LFM-BPSK Composite Modulated Radar Signal Based on Multi-Output Regression Network

Aiming at the problem of high relative error in parameter estimation of LFM-BPSK radar signal by traditional decomposition algorithm, this paper proposes a parameter estimation algorithm for LFM-BPSK based on Multi-Output Regression Network (MORN). The algorithm first innovatively designs the ID-CNN-BiGRU network pre-training for feature extraction, then weights migration to the MORN network. The simulation results show that the relative errors of symbol rate, modulation frequency slope and carrier frequency are as low as 9.73 %, 10.80 % and 6.65 % at −6dB SNR respectively. The algorithm has good engineering value and can be extended to more composite modulated radar signals.