Michael D. Collins
W. A. Kuperman
J. S. Perkins
Naval Res. Lab., Washington, DC 20375
(Angstrom)ge Kristensen
SACLANT Undersea Res. Ctr., La Spezia, Italy
It is impossible to localize an acoustic source in the ocean with conventional matched-field processing techniques if there are significant uncertainties in the acoustic parameters of the environment. Focalization is an approach for overcoming this difficulty that involves searching for key environmental parameters in addition to searching for the source location [Collins and Kuperman, J. Acoust. Soc. Am. 90, 1410--1422 (1991)]. The feasibility of this approach was investigated using implementations that might be difficult to apply to data. An implementation of focalization involving the parabolic equation method for computing replica fields and eigenvector-based signal processing is applicable to problems involving data from complex ocean environments. The covariance matrix is formed for a frequency corresponding to a peak in the source spectrum. The input data to the processor is the eigenvector corresponding to the largest eigenvalue. A large signal-to-noise gain is achieved with this approach, and the eigenvector is often a good representation of the true source field. To determine the source location, the eigenvector is compared with replica fields using a cost function that is minimized over the environmental parameter space with simulated annealing.