Minimum variance distortionless response multi-source resolving using inhomogeneous diagonal unloading

Under some underwater situations, the multi-source resolving performance of the Minimum Variance Distortionless Response (MVDR) method could seriously degrade due to the strong and inhomogeneous uncorrelated noise on the hydrophone array. To mitigate this problem, we propose an Inhomogeneous Diagonal Unloading MVDR (IDUMVDR) method, where inhomogeneous diagonal unloading is applied on the covariance matrix before applying the MVDR method. The diagonal unloading level for each array element is obtained by solving a positive semi-definite optimization problem, where we maximize the sum of the diagonal unloading levels while constraining the minimum eigenvalue of the unloaded covariance matrix to be a small positive value. The numerical simulations suggested that the proposed IDU-MVDR method can remove most of the uncorrelated noise while remaining a small amount of which on the array elements. In consequence, the IDU-MVDR method provides higher resolution, lower background level, more distinct peaks and guaranteed robustness compared to the MVDR method. The sea trial results agreed well with the numerical simulations and proved the vitality of the IDU-MVDR method.