A Single-channel Blind Source Separation Method based on Modulation Parameter Estimation and Signal Reconstruction

In modern warfare, the electromagnetic spectrum control becomes one of the key factors that determines the outcome of battles. Blind source separation (BSS) technology has been widely applied in electronic reconnaissance to effectively separate various source signals. Addressing the poor performance of traditional single-channel BSS (SCBSS) algorithms under low signal-to-noise ratio (SNR) conditions, this paper proposes a SCBSS algorithm based on modulation parameter estimation and demodulated short-time Fourier transform (DSTFT) reconstruction. First, the parameter estimation problem is formulated as an optimization problem, and the instantaneous frequency (IF) and phase of the source signals are accurately estimated through using the quantum-inspired bald eagle search (QBES) algorithm. Then, the DSTFT of the mixed signal is performed to obtain the enhanced time-frequency (TF) distribution of the corresponding source signal. Finally, the source signals are precisely reconstructed by utilizing the TF coefficients near the estimated IF trajectory. Through extensive simulation experiments, the effectiveness of the proposed algorithm in signal separation is validated, particularly under low SNR conditions, where it demonstrates superior noise robustness compared to the competing algorithms.