No abstract

This paper investigates the problem of cancellation of noise generated by own platform in shallow water scenario. In the case of underwater acoustics, the target signal detection and tracking in the presence of tow ship noise is a challenging task. A computationally intensive technique is necessary for tow ship noise suppression. In this paper, an algorithm using deep regression neural network (DRNN) along with minimum variance distortionless response (MVDR) beamformer is presented for tow ship noise cancellation. Nine DRNN’s each with different weight initialization techniques and activation functions are designed for effective tow ship noise cancellation. The designed DRNNs is tested using the simulated data and further validated using the real data collected during the trials from Arabian Sea.

In this paper, we attempt to unify two array processing frameworks viz, Acoustic Vector Sensor (AVS) and two level nested array to enhance the Degrees of Freedom (DoF) significantly beyond the limit that is attained by a Uniform Linear Hydrophone Array (ULA) with specified number of sensors. The major focus is to design a line array architecture which provides high resolution unambiguous bearing estimation with increased number of spatial nulls to mitigate the multiple interferences in a deep ocean scenario. AVS can provide more information about the propagating acoustic field intensity vector by simultaneously measuring the acoustic pressure along with tri-axial particle velocity components. In this work, we have developed Nested AVS array (NAVS) ocean data model to demonstrate the performance enhancement. Conventional and MVDR spatial filters are used as the response function to evaluate the performance of the proposed architecture. Simulation results show significant improvement in performance viz, increase of DoF, and localization of more number of acoustic sources and high resolution bearing estimation with reduced side lobe level.

No abstract