Performance analysis of model-based localization of high-frequency acoustic sources in 3D

A robust model-based source localization (MBSL) scheme for 3D positioning of high-frequency underwater acoustic (UWA) transmitters, as well as performance assessment of 3D speed vector estimation of the acoustic source, deploying two linear receiver arrays, is developed and evaluated. In addition, this paper presents the sensitivity evaluation of MBSL to uncertainties involved in environmental parameters, including the effects of sound speed profile and bathymetry mismatches. The results presented here pertain to the general framework of inferring side information concerning an UWA link from the environmental signature imprinted on waveforms by the propagation medium. The presented algorithms directly operate on estimated linear time-varying UWA channel responses, represented as 3-dimensional delay-Doppler-depth functions (DDDF). Estimation of DDDF coefficients is achieved jointly for all receivers deployed at different depths using Basis-Pursuit (BP) tools, which can efficiently handle sparse unconstrained ℓ
₂
- ℓ
₁
minimization, and directly operate on the complex signals of baseband models. To extract delay/Doppler information related to each path from the DDDF representation, an efficient energy-based scheme is developed to detect and label wavefronts using clustering techniques inspired by image processing.