Robust DOA-Aided Sensor-Array Receiver for Mobile Underwater Acoustic Communications Under Directional Impulsive Noise

Abstract

Underwater acoustic sensor-array receivers deployed in under-ice mobile communication scenarios are challenged by directional impulsive noise and path-dependent Doppler distortion. This article proposes a robust direction-of-arrival (DOA)-aided sensor-array receiving framework that integrates broadband beamforming and low-complexity multichannel decision feedback equalization (MDFE). At the array-processing front end, a normalized-kurtosis-ratio (NKR) frame weighting scheme is embedded into coherently focused wideband MUSIC to down-weight impulsively contaminated array snapshots and suppress nonpath DOA pseudo-peaks. The estimated multipath DOAs are then used to construct frequency-domain null-steering beamformers, which transform the element-domain array measurements into path-dominated branches and prevent noncommunication directions from entering the back-end receiver. At the back end, each branch is processed by symbol-level real-time resampling and phase compensation to correct path-specific Doppler, followed by MDFE-based multipath combining. A real-time-resampling-compatible dichotomous-coordinate-descent (DCD) recursive least squares (RLSs) update with block-wise recursive correlation updates is introduced to avoid explicit matrix inversion and reduce adaptive-filtering complexity. Simulations with measured under-ice impulsive noise evaluate the receiver under rapidly time-varying path-dependent Doppler, showing the lowest observed BER and a 41.7% runtime reduction compared with conventional real-time resampling MDFE. Songhua Lake trial data with superimposed measured impulsive noise further validate robust DOA estimation, path separation, residual Doppler tracking, decoding reliability, and runtime reduction under slow source motion and real-channel propagation, achieving the lowest average BER of 0.0020 and the shortest average runtime of 1.889 s among the compared methods.