Kathleen E. Wage
Arthur B. Baggeroer
Res. Lab. of Electron., MIT, 77 Massachusetts Ave., 36-615, Cambridge, MA 02139
The normal mode amplitudes and their second-order statistics are useful in understanding sound propagation in the medium and for matched-mode processing (MMP) and matched-field tomography (MFT). Standard modal beamforming techniques introduce a least-squares error criterion to compute the modal amplitude estimates. This work uses a method of adaptive estimation which is capable of processing both coherent and incoherent modes. The new approach is fundamentally different from other modal estimators (e.g., MMP) because it is data-adaptive and maximizes SNR against an ambient noise background instead of minimizing squared error in the estimate without regard to the noise. The new methods are studied using simulations which include coherent and incoherent modes and a realistic ocean noise model. The performance of the estimators is evaluated with respect to the following criteria: (i) orthogonality of the sampled mode shapes, (ii) power level and spatial structure of the noise, (iii) mode coherence, (iv) modeling assumptions (number of modes to estimate), and (v) the presence of multiple modal sources. The new methods perform significantly better than least squares in high noise environments and in situations where the sampled modes are not orthogonal.