Adaptive enhancer of the target radiated line-spectrum under non-Gaussian noise

The signal-to-noise ratio (SNR) of the underwater target radiated line-spectrum cannot be effectively improved by conventional adaptive line enhancer under the presence of non-Gaussian noise, as the weight iteration diverges due to the divergence of the noise second-order moment. An adaptive line enhancer based on the maximum correntropy criterion is proposed to address this problem. The correntropy criterion between the reference signal and output signal is used as a cost function for iterative optimization of adaptive filter weights, taking advantage of the difference between narrowband line-spectrum and wideband non-Gaussian noise in the generalized time correlation radius, which effectively improves the output SNR of target radiated under non-Gaussian noise. Simulation results show that the proposed method enhances line-spectrum signals under non-Gaussian noise, and outperforms two other adaptive line enhancer methods based on the singular spectral analysis and Shannon entropy under non-Gaussian noise, respectively improving the SNR by 12 dB and 5 dB in low SNR. The feasibility and reliability of the proposed method are confirmed by results from sea trial data processing.