UC San Diego

Passive acoustic signal processing has the remarkable capacity to achieve traditional goals of active sonar, such as detection and ranging, simply by listening. Matched field processing (MFP) exploits the physics of wave propagation to design filters tailored to the underwater acoustic environment and solve passive sonar problems of detection and localization. The second chapter of this dissertation introduces range-coherent MFP to increase the coherent integration time of the signal from a moving source. Range-coherent MFP searches over candidate source tracks to compute time-dependent replicas that are then used to coherently combine data over a vertical array and over multiple resolution cells.The method improves on existing methods by extending the application of MFP to low signal-to-noise ratio (SNR) regimes. The third chapter introduces modal-MUSIC to estimate waveguide properties by using the structure of noise recorded on a vertical line array. Modal-MUSIC estimates the horizontal wavenumbers of the modal field from noise on a partially-spanning array. It relaxes the requirement of state-of-the art passive methods which require a full water column spanning array. In this chapter, the modal field parameters estimated from noise with modal-MUSIC are successfully used to localize a source at high SNR without an environmental model. In the fourth chapter, these two developments are brought together to demonstrate the feasibility of MFP at low SNR in the absence of extensive \textit{a priori} information on the seabed properties. The fourth chapter improves upon the model-free localization results with modal-MUSIC in chapter three by fitting the modal wavenumbers to a geoacoustic model. This dissertation extends the application of MFP to the passive localization of a low SNR moving source in the absence of knowledge of the seabed properties.