Speed-up and detection performance of a nearest-neighbors algorithm for matched-field processing with a vertical line array

Matched-field processing (MFP) matches measured acoustic fields of an underwater acoustic source against a full set of replica fields filling a volume of ocean in which the detection and localization of the source is desired. Selecting for matching those replica fields that are most like the data, i.e., the nearest-neighbors (NNs) to the data, offers a way of reducing the computational search space. By applying this technique an order of magnitude speedup was obtained. The probability of detection for the Bartlett beamformer, with and without selection of the replicas, was estimated from simulations of noisy data received on a vertical line array at practical time-bandwidth products. To achieve the same detection performance over the search region for four detection criteria with selection required signal-to-noise ratios between 0.2 and 2.0 dB higher than without selection for the same false alarm probability