A test of autonomous matched-field tracking with a sparse seafloor array

A recent experiment demonstrated the feasibility of using input from a sparse seafloor array of hydrophones [Stevenson and Bucker, J. Acoust. Soc. Am. 108, 2645 (2000)] into a matched-field tracking algorithm [Bucker, J. Acoust. Soc. Am. 96, 3809–3811 (1994)] to return tracks of submerged sound sources. In previous work, the algorithm was tested using recorded data. This paper describes a near real-time implementation of this processing. The tracking algorithm operates in near real-time in a DSP (digital signal processor) running autonomously inside the unit. The procedure is to select N Fourier frequency bins based on S/N ratios after CFAR normalization, and then to determine the best tracks using M-of-N processing. Observations over 3–6 min tracks are matched against the calculated sound field stored in a look-up table [Bucker and Baxley, J. Acoust. Soc. Am. 106, 3226–3230 (1999)]. We compare near real-time results obtained autonomously with those obtained in post-experiment analysis. [Work supported by ONR, Code 321-SS.]