Adaptive Genetic Mode Decomposition Algorithm for Single-Channel Dense Multisignal Mix Signal Separation in Optical Frequency Comb-Based Down-Conversion Wideband Signal Measurement

In optical frequency comb (OFC) undersampling, the downconverted signal spectrum exhibits high spectral density, and traditional single-channel mix signal separation (SC-MSS) algorithms often fail to meet the requirements. To address these challenges, an adaptive genetic mode decomposition (GMD) algorithm is proposed, achieving high-precision adaptive separation of single-channel mixed signals in complex electromagnetic environments. The algorithm initializes the target signal for separation into several genetic modes. The key parameters, such as the number of modes, signal bandwidths, center frequencies, amplitudes, phases, and window functions, are encoded into gene strings (chromosomes). The optimization objectives include minimizing signal reconstruction error, minimizing total bandwidth, and minimizing mode overlap. A genetic population is then constructed to iteratively evolve and optimize the GMD. The simulation results demonstrate that, under a signal-to-noise ratio (SNR) of 0 dB, the proposed GMD algorithm attains an average correlation coefficient of 0.98, an average signal-to-distortion ratio (SDR) of 14.16 dB, and an average signal-to-interference ratio (SIR) of 85.34 dB. Compared to variational mode decomposition (VMD), these metrics represent improvements of 42%, 13.78 dB, and 52.79 dB, respectively, all significantly outperforming the traditional VMD algorithm. Across various signal types, the maximum frequency resolution of the GMD algorithm reaches 100 kHz, with an average resolution enhancement of approximately 80% over VMD. In terms of processing speed, GMD operates at only 11.3% of the runtime required by VMD for the same number of signals.