Target motion analysis in the deep-sea shadow zone based on surface waveguide leakage

Passive tracking of underwater targets in the deep-sea acoustic shadow zone is hindered by biased angle measurements, limited receiver maneuverability, and large measurement fluctuations. To address these difficulties, this paper develops a physics-based horizontal-line-array measurement model that combines the spatial angles of the surface waveguide leakage and bottom-reflected signals and incorporates it into a particle filter for highly nonlinear state estimation. Simulations and sea-trial data demonstrate tracking without receiver maneuvering and robustness to intermittent loss of the bottom-reflected observation. Across loss rates from 0 to 0.95, the particle filter reduces the position root mean square error by 50.4%-72.3% relative to the best result among extended, unscented, and cubature Kalman filters. A sea trial turning-point analysis gives a one-step prediction root mean square error of 12.14 m in turning segments, compared with 2.33 m in straight segments; measurement updates largely correct this local mismatch when bottom-reflected observations are available. These results quantify both the benefit of the non-Gaussian particle representation and the limitation of the adopted piecewise-constant-velocity prior during turns with incomplete range-related observations.