A Separation Algorithm for Satellite-Based AIS Signals Based on Deep Learning

In the dense area of inland river ships, the collision of ship automatic identification system (AIS) signals often occurs. This collision problem affects obstacle avoidance decision-making. Aiming at the above issues, this paper proposes a deep learning-based signal blind source separation processing method. By constructing a deep learning architecture composed of denoising self-encoder and deep neural network, the downscaled features of the input signal are extracted and reconstructed into the original signal, and through the learning and training of a large amount of signal data, the model can independently extract the signal features and mixing laws, to realize the effective denoising and separation of the original AIS signal. Simulation results show that the Pearson correlation coefficient of this method maintains at 0.885\~{}0.981 under different signal-to-noise ratios (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$5 \sim 30 ~\text{dB}$</tex>), which is better than some traditional algorithms for denoising and signal separation, and provides reliable AIS information support for obstacle avoidance, which is of great significance in enhancing the safety of ship navigation.