Separation of Communication Signals Based on Underdetermined Blind Source Separation

The paper proposes an improved underdetermined blind source separation (UBSS) method of nonlinear mixture, which can be applied to the communication signals' separation during data link transmission in the complicated electromagnetic environment. The proposed method combines the advantages of wavelet analysis and kernel canonical correlation analysis (KCCA), and can enhance the signals' detection performance to the data link communication via separating the communication signals from the aliasing signals effectively. The simulation results show that the wavelet-KCCA BSS method can pick up the objective communicative signals even in low SNR condition, which has proper feasibility and effectiveness.