Passive Sonar Array Sub-space Processing based on Modal Decomposition

In passive underwater sonar, current attention is on large arrays which can provide the needed array gain and high resolution beamforming. However, full-rank adaptive processing for these arrays requires sufficient sample support which if often difficult to acquire in dynamic shallow water environments. Furthermore, complicated acoustic propagation and uncertainties about the ocean parameters can limit the effectiveness of adaptive techniques. Several researchers have proposed mode-based processing methods for Matched Field Processing, depth-dependent pre-filtering, and reduced rank processing. For arrays that provide the ability to decompose the field into modes, these methods are believed to be less sensitive to environmental mismatch with the added potential advantage of computational efficiency. In this paper, we attempt to quantify the performance of mode-based processing in uncertain environments, and compare this performance to more classic processing approaches. Results are presented from Monte Carlo simulations of range-dependent environments in which the uncertainty in the sound speed and bottom bathymetry is varied parametrically.