Source number estimation in shallow ocean by acoustic vector sensor array using Gerschgorin disks

Accurate estimation of the number of sources plays a crucial role in the source localization and direction-of-arrival estimation problems. In several applications like coastal surveillance and harbour defence, source number estimation has to be performed in a shallow ocean environment. In this paper, the Gerschgorin disk estimator (GDE) method of source number estimation in an unbounded medium is extended for operation in a shallow oceanic waveguide. Classical methods of source number estimation such as Akaike information criterion (AIC) and minimum description length (MDL) require (1) good estimates of the eigenvalues of the spectral correlation matrix, and (2) the assumption of white Gaussian noise to be valid. The GDE method does not suffer from these limitations. A theoretical formulation of the GDE method in a shallow ocean is presented in this paper for acoustic vector sensor (AVS) and acoustic pressure sensor (APS) arrays. Simulation results are then presented to illustrate the advantages of the GDE method and the superior performance of the AVS array in a shallow ocean environment.