A method of weak signal detection based on subband ANC and normalized energy spectrum

In order to improve the weak signal detection performance of Unmanned Underwater Vehicle (UUV), depressing its self-generated noise and fully using the differences of incidence in space domain and differences in frequency domain (spectral patterns and spectral level) between signal and noise are two key issues. In practice, there are many measures besides ANC (adaptive noise cancellation) to enhance the SNR of output of element. In this paper, a method of weak signal detection was proposed, where in order to improve the performance of ANC we selected multi-channel references to enhance the performance of coherence, and non-uniform subband processing to increase the performance of convergence speed in each output of element; in space domain processing, normalized energy spectrum and beam forming in frequency domain was used. The results of simulation and lake experiments proved the validity of the method.