Early Interference Detection in HFSWR Input Channel using Convolutional Neural Network

Abstract

Early detection of ionospheric interference is crucial for continuous monitoring of remote sea areas of exclusive economic zone (EEZ) using high-frequency over the horizon radar (HF-OTHR). In this paper, approach with convolutional neural networks and transfer learning is proposed for detection of the regions affected with ionospheric and other types of interference instead of conventional spectrum analysis. Early detection of interference that is frequency dependent may lead to changing the operating frequency of radar. Adopted approach provided very good results.