Adaptive Receiver Based on Maximum Correntropy Criterion Robust to Non-Gaussian Noise for In-Band Full-Duplex Underwater Communications

Abstract

To enhance spectral efficiency in underwater communications, in-band full-duplex transmission has emerged as a promising solution. However, the presence of non-Gaussian noise in underwater channels poses significant challenges for conventional self-interference cancellation and channel equalization algorithms. In this paper, we propose self-interference cancellation and channel equalization algorithms based on the maximum correntropy criterion, a machine learning technique rooted in information theory. These algorithms effectively mitigate non-Gaussian noise, leading to improved system performance. Simulation results demonstrate a 4dB improvement in signal-to-noise ratio (SNR) over conventional RLS-based methods under non-Gaussian noise conditions, significantly reducing the bit error rate (BER) and enhancing the reliability of underwater communications. The proposed receiver achieves robust performance across diverse noise models, validating its effectiveness in practical underwater environments.