Implication of shallow waters for source localization: Time-delays estimation versus matched-field processing

Since matched-field processing (MFP) has been widely studied, and due to the increasing interest about shallow water conditions, this paper addresses the rightful problem of comparing such a promising technique with the classical multipath time-delay estimation (TDE) and least-squares fitting, for a source localization purpose. At first the response function of both methods (MFP and TDE) are established for a single broadband source, considering an equispaced horizontal linear array and taking into account the assignment ambiguity between the set of time-delay measurements and all the possible model replicas. Then their Cramer-Rao bounds are concurrently processed and compared in the case where the acoustic medium is perfectly known. Finally, biases in range and depth estimation are studied for TDE and a robust version of MFP (ie. adaptive with multiple constraints) in the presence of environmental mismatch as celerity gradient errors. TDE reveals itself as being more robust and less sensitive than MFP.