Denoising of Over-the-Horizon Propagation Loss Based on 1D Convolutional Autoencoder

Over-the-horizon radar systems have attracted widespread attention because they target at resolving the challenges of marine communication over long distance and weakened signals. However, these systems are limited by signal noise. To address this issue, we construct a denoising model based on a one-dimensional convolutional autoencoder (1DCAE) to improve the quality of the input data. Experiments show that the root mean square and mean absolute errors of the proposed 1DCAE are dramatically reduced, compared with those of the model without using the 1DCAE filter. Our proposed 1DCAE deep learning framework outperforms the other state-of-the-art techniques in terms of over-the-horizon propagation loss signal noise.