Self-Calibrated Compound-Gaussian Regularized Adaptive Beamforming for Robust Ocean Remote Sensing

Abstract

Ocean remote sensing is essential for marine surveillance, ship detection, environmental monitoring, offshore safety, and underwater target observation. However, reliable sensing in ocean environments remains challenging because received signals are affected by sea clutter, reverberation, multipath propagation, and low signal-to-noise ratio. In practical radar and sonar systems, sea clutter is often non-Gaussian and varies with sea state, reducing the reliability of conventional beamforming methods. Classical beamforming suffers from weak interference suppression and high sidelobes, while adaptive methods such as MVDR are sensitive to covariance estimation errors, limited snapshots, and steering mismatch, which may cause target self-nulling. To address these limitations, this research proposes a self-calibrated compound-Gaussian regularized adaptive beamforming with waveform-constrained sidelobe optimization (SC-CG-RABF) framework for robust ocean remote sensing. The main contribution is not the separate use of compound-Gaussian clutter modeling, diagonal loading, or PSLR-constrained waveform design, since these methods are already established. Instead, the novelty lies in integrating target-protected texture-normalized covariance calibration, regularized adaptive beamforming, and waveform-constrained sidelobe control into one unified processing framework. Here, self-calibration refers to data-driven covariance calibration using target-protected compound-Gaussian texture estimation, not hardware or feedback-based calibration. Compared with conventional beamforming, MVDR, and regularized MVDR, the proposed SC-CG-RABF jointly addresses non-Gaussian sea clutter, limited snapshots, covariance mismatch, steering-vector sensitivity, interference leakage, and waveform sidelobe effects. The results show a narrower and more stable main lobe, lower sidelobes, deeper interference nulls, and higher output SINR. The waveform design also achieves the best PSLR of 17.91 dB. Using a real Sentinel-1 SAR ocean patch, the method improves SCR from 1.17 to 1.38 dB, increases local contrast from 3.3502 to 5.1444, and reduces clutter power by 1.04 dB.