Spatial-Time-Frequency Information Collaborative Learning for Ionosphere Clutter Suppression in HFSWR

Abstract

Eliminating the ionosphere clutter is particularly important for high frequency surface wave radar (HFSWR). This article fills the gap between the growing interests in the application of deep learning network and the new approach of ionosphere clutter suppression. We present a dynamic collaborative learning strategy to simultaneously learn the spatial-time-frequency (STF) information of multi-component radar echoes for suppressing the ionosphere clutter. For capturing the multi-dimension information, we design a multi-channel time-frequency characteristic learning network (MTF) and a multi-channel spatial characteristic learning network (MS) to sufficiently learn the characteristics of multi-component radar echoes according to the divided regions. The proposed methodology not only suppresses the ionosphere clutter but also preserves the targets from contaminated region after the suppression process. The results presented in the letter allow demonstrating how the proposed methodology outperforms the approaches taken as reference in all the cases under study.