UC San Diego

Localizing and tracking underwater targets is of interest to the study of sound emitting marine life, physical or chemical phenomena affecting acoustic wave's propagation or to naval SONARS, to cite a few applications. Matched- field processing (MFP) is a passive localization method based on comparing the pressure fields received on an hydrophone array with synthesized fields coming from hypothetical source locations. The best match between modeled and received signals yields an estimate of the source's position. Successful localization requires a number of conditions to be met. Low-frequency, loud signals propagating in a steady well studied environment without presence of loud interferers have been tracked over tens to hundreds of kilometers in the literature. This dissertation addresses MFP's localization and tracking performance at low signal-to-noise ratios, in the presence of environmental mismatch between modeled and real propagating environments and/or for high frequency scenarios. The issue of robustness and snapshot deficiency are addressed by implementing the white noise constraint adaptive processor. Throughout most of the dissertation, a comparison between conventional, minimum variance distortionless response and white noise constraint algorithms' localization performance is provided. Broadband algorithms, processing frequencies both in a coherent and incoherent way are introduced to enhance MFP's ability to detect low signal-to-noise ratio sources. A technique to estimate and include the unknown source phase to the processing is developed, and the extra-gain it provides is determined. The enhanced sensitivity of MFP to environmental variability in high-frequency scenarios is investigated using data emitted from multiple sources to create replica vectors. A combination of robust adaptive MFP using the white noise constraint method and coherent broadband processing was shown to yield promising localization results in high-frequency scenarios for which MFP is typically problematic