Mitigating ambient noise in underwater acoustic receivers using independent component analysis

The acoustic ambience of the ocean is extremely turbid with a myriad of heterogeneous competing sources, of both manmade and natural origin. The cumulative clamour of sources like shipping, sonar, coastal machines, marine life, shore waves, as well as other hydrodynamic activities can often lift the noise floor to much higher levels so that the vital signals are masked and buried well below the recognizable level. The accuracy and reliability of acoustic systems and devices depend on the degree of observability of a signal above the background noise or the ambient noise floor. In this paper an attempt is made to reveal the latent source signals from the observed mixture by carrying out the Independent Component Analysis (ICA) thereby mitigating the ambient noise levels to a practically acceptable level. In the proposed prototype, FastICA, the most popular and computationally efficient algorithm is being utilized.